diff --git a/src/diffusers/modular_pipelines/flux/before_denoise.py b/src/diffusers/modular_pipelines/flux/before_denoise.py index 4272066309a2..1b6e16a940bb 100644 --- a/src/diffusers/modular_pipelines/flux/before_denoise.py +++ b/src/diffusers/modular_pipelines/flux/before_denoise.py @@ -12,13 +12,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect -from typing import Any, List, Optional, Tuple, Union +from typing import Any, List, Optional, Tuple import numpy as np import torch from ...models import AutoencoderKL +from ...pipelines.pipeline_utils import calculate_shift, retrieve_latents, retrieve_timesteps from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import logging from ...utils.torch_utils import randn_tensor @@ -30,80 +30,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - # Adapted from the original implementation. def prepare_latents_img2img( vae, scheduler, image, timestep, batch_size, num_channels_latents, height, width, dtype, device, generator @@ -146,20 +72,6 @@ def prepare_latents_img2img( return latents, latent_image_ids -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def _pack_latents(latents, batch_size, num_channels_latents, height, width): latents = latents.view(batch_size, num_channels_latents, height // 2, 2, width // 2, 2) latents = latents.permute(0, 2, 4, 1, 3, 5) diff --git a/src/diffusers/modular_pipelines/flux/encoders.py b/src/diffusers/modular_pipelines/flux/encoders.py index 8c49990280ac..43fd41910845 100644 --- a/src/diffusers/modular_pipelines/flux/encoders.py +++ b/src/diffusers/modular_pipelines/flux/encoders.py @@ -23,6 +23,7 @@ from ...image_processor import VaeImageProcessor from ...loaders import FluxLoraLoaderMixin, TextualInversionLoaderMixin from ...models import AutoencoderKL +from ...pipelines.pipeline_utils import retrieve_latents from ...utils import USE_PEFT_BACKEND, is_ftfy_available, logging, scale_lora_layers, unscale_lora_layers from ..modular_pipeline import ModularPipelineBlocks, PipelineState from ..modular_pipeline_utils import ComponentSpec, ConfigSpec, InputParam, OutputParam @@ -53,20 +54,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class FluxVaeEncoderStep(ModularPipelineBlocks): model_name = "flux" diff --git a/src/diffusers/modular_pipelines/qwenimage/before_denoise.py b/src/diffusers/modular_pipelines/qwenimage/before_denoise.py index 606236cfe91b..c370157d9c8e 100644 --- a/src/diffusers/modular_pipelines/qwenimage/before_denoise.py +++ b/src/diffusers/modular_pipelines/qwenimage/before_denoise.py @@ -12,13 +12,13 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect -from typing import List, Optional, Tuple, Union +from typing import List, Tuple import numpy as np import torch from ...models import QwenImageControlNetModel, QwenImageMultiControlNetModel +from ...pipelines.pipeline_utils import calculate_shift, retrieve_timesteps from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils.torch_utils import randn_tensor, unwrap_module from ..modular_pipeline import ModularPipelineBlocks, PipelineState @@ -26,80 +26,6 @@ from .modular_pipeline import QwenImageModularPipeline, QwenImagePachifier -# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - # modified from diffusers.pipelines.stable_diffusion_3.pipeline_stable_diffusion_3_img2img.StableDiffusion3Img2ImgPipeline.get_timesteps def get_timesteps(scheduler, num_inference_steps, strength): # get the original timestep using init_timestep diff --git a/src/diffusers/modular_pipelines/qwenimage/encoders.py b/src/diffusers/modular_pipelines/qwenimage/encoders.py index 2ab83a03ee55..a1382e84aa3f 100644 --- a/src/diffusers/modular_pipelines/qwenimage/encoders.py +++ b/src/diffusers/modular_pipelines/qwenimage/encoders.py @@ -22,6 +22,7 @@ from ...guiders import ClassifierFreeGuidance from ...image_processor import InpaintProcessor, VaeImageProcessor, is_valid_image, is_valid_image_imagelist from ...models import AutoencoderKLQwenImage, QwenImageControlNetModel, QwenImageMultiControlNetModel +from ...pipelines.pipeline_utils import retrieve_latents from ...pipelines.qwenimage.pipeline_qwenimage_edit import calculate_dimensions from ...utils import logging from ...utils.torch_utils import unwrap_module @@ -128,20 +129,6 @@ def get_qwen_prompt_embeds_edit( return prompt_embeds, encoder_attention_mask -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Modified from diffusers.pipelines.qwenimage.pipeline_qwenimage.QwenImagePipeline._encode_vae_image def encode_vae_image( image: torch.Tensor, diff --git a/src/diffusers/modular_pipelines/stable_diffusion_xl/before_denoise.py b/src/diffusers/modular_pipelines/stable_diffusion_xl/before_denoise.py index 70cbf0c1c78d..1d7740aedf27 100644 --- a/src/diffusers/modular_pipelines/stable_diffusion_xl/before_denoise.py +++ b/src/diffusers/modular_pipelines/stable_diffusion_xl/before_denoise.py @@ -12,8 +12,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect -from typing import Any, List, Optional, Tuple, Union +from typing import Any, List, Optional, Tuple import PIL import torch @@ -23,6 +22,7 @@ from ...image_processor import VaeImageProcessor from ...models import AutoencoderKL, ControlNetModel, ControlNetUnionModel, UNet2DConditionModel from ...models.controlnets.multicontrolnet import MultiControlNetModel +from ...pipelines.pipeline_utils import retrieve_latents, retrieve_timesteps from ...schedulers import EulerDiscreteScheduler from ...utils import logging from ...utils.torch_utils import randn_tensor, unwrap_module @@ -43,80 +43,6 @@ # configuration of guider is. -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def prepare_latents_img2img( vae, scheduler, image, timestep, batch_size, num_images_per_prompt, dtype, device, generator=None, add_noise=True ): diff --git a/src/diffusers/modular_pipelines/stable_diffusion_xl/encoders.py b/src/diffusers/modular_pipelines/stable_diffusion_xl/encoders.py index 90b254b6f5d4..da112cf78077 100644 --- a/src/diffusers/modular_pipelines/stable_diffusion_xl/encoders.py +++ b/src/diffusers/modular_pipelines/stable_diffusion_xl/encoders.py @@ -29,6 +29,7 @@ from ...loaders import StableDiffusionXLLoraLoaderMixin, TextualInversionLoaderMixin from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel from ...models.lora import adjust_lora_scale_text_encoder +from ...pipelines.pipeline_utils import retrieve_latents from ...utils import ( USE_PEFT_BACKEND, logging, @@ -43,20 +44,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionXLIPAdapterStep(ModularPipelineBlocks): model_name = "stable-diffusion-xl" diff --git a/src/diffusers/modular_pipelines/wan/before_denoise.py b/src/diffusers/modular_pipelines/wan/before_denoise.py index d48f678edd59..7a19425162e1 100644 --- a/src/diffusers/modular_pipelines/wan/before_denoise.py +++ b/src/diffusers/modular_pipelines/wan/before_denoise.py @@ -12,11 +12,11 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import List, Optional, Union import torch +from ...pipelines.pipeline_utils import retrieve_timesteps from ...schedulers import UniPCMultistepScheduler from ...utils import logging from ...utils.torch_utils import randn_tensor @@ -34,66 +34,6 @@ # configuration of guider is. -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class WanInputStep(ModularPipelineBlocks): model_name = "wan" diff --git a/src/diffusers/pipelines/allegro/pipeline_allegro.py b/src/diffusers/pipelines/allegro/pipeline_allegro.py index 3be0129088fb..06411f11250c 100644 --- a/src/diffusers/pipelines/allegro/pipeline_allegro.py +++ b/src/diffusers/pipelines/allegro/pipeline_allegro.py @@ -39,6 +39,7 @@ ) from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import AllegroPipelineOutput @@ -81,66 +82,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AllegroPipeline(DiffusionPipeline): r""" Pipeline for text-to-video generation using Allegro. diff --git a/src/diffusers/pipelines/animatediff/pipeline_animatediff_sdxl.py b/src/diffusers/pipelines/animatediff/pipeline_animatediff_sdxl.py index 56d319027595..dc9363839e9c 100644 --- a/src/diffusers/pipelines/animatediff/pipeline_animatediff_sdxl.py +++ b/src/diffusers/pipelines/animatediff/pipeline_animatediff_sdxl.py @@ -57,7 +57,7 @@ from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor from ..free_init_utils import FreeInitMixin -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from .pipeline_output import AnimateDiffPipelineOutput @@ -147,66 +147,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AnimateDiffSDXLPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/animatediff/pipeline_animatediff_sparsectrl.py b/src/diffusers/pipelines/animatediff/pipeline_animatediff_sparsectrl.py index 46d650efe8b6..4409584a3d2d 100644 --- a/src/diffusers/pipelines/animatediff/pipeline_animatediff_sparsectrl.py +++ b/src/diffusers/pipelines/animatediff/pipeline_animatediff_sparsectrl.py @@ -39,7 +39,7 @@ from ...utils.torch_utils import is_compiled_module, randn_tensor from ...video_processor import VideoProcessor from ..free_init_utils import FreeInitMixin -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from .pipeline_output import AnimateDiffPipelineOutput @@ -115,20 +115,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class AnimateDiffSparseControlNetPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video.py b/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video.py index 6f3a609aba4a..fe2f97ce8b47 100644 --- a/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video.py +++ b/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video.py @@ -36,7 +36,7 @@ from ...video_processor import VideoProcessor from ..free_init_utils import FreeInitMixin from ..free_noise_utils import AnimateDiffFreeNoiseMixin -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from .pipeline_output import AnimateDiffPipelineOutput @@ -104,80 +104,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AnimateDiffVideoToVideoPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video_controlnet.py b/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video_controlnet.py index b00f344598ad..3271c6a37d6e 100644 --- a/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video_controlnet.py +++ b/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video_controlnet.py @@ -44,7 +44,7 @@ from ...video_processor import VideoProcessor from ..free_init_utils import FreeInitMixin from ..free_noise_utils import AnimateDiffFreeNoiseMixin -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from .pipeline_output import AnimateDiffPipelineOutput @@ -122,80 +122,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AnimateDiffVideoToVideoControlNetPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/aura_flow/pipeline_aura_flow.py b/src/diffusers/pipelines/aura_flow/pipeline_aura_flow.py index 6251ca443533..981fd2f5b77f 100644 --- a/src/diffusers/pipelines/aura_flow/pipeline_aura_flow.py +++ b/src/diffusers/pipelines/aura_flow/pipeline_aura_flow.py @@ -11,7 +11,6 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import torch @@ -32,7 +31,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_timesteps if is_torch_xla_available(): @@ -60,66 +59,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AuraFlowPipeline(DiffusionPipeline, AuraFlowLoraLoaderMixin): r""" Args: diff --git a/src/diffusers/pipelines/bria/pipeline_bria.py b/src/diffusers/pipelines/bria/pipeline_bria.py index ebddfb0c0eee..74f177a13c26 100644 --- a/src/diffusers/pipelines/bria/pipeline_bria.py +++ b/src/diffusers/pipelines/bria/pipeline_bria.py @@ -15,7 +15,6 @@ from ...models.transformers.transformer_bria import BriaTransformer2DModel from ...pipelines import DiffusionPipeline from ...pipelines.bria.pipeline_output import BriaPipelineOutput -from ...pipelines.flux.pipeline_flux import calculate_shift, retrieve_timesteps from ...schedulers import ( DDIMScheduler, EulerAncestralDiscreteScheduler, @@ -31,6 +30,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor +from ..pipeline_utils import calculate_shift, retrieve_timesteps if is_torch_xla_available(): diff --git a/src/diffusers/pipelines/chroma/pipeline_chroma.py b/src/diffusers/pipelines/chroma/pipeline_chroma.py index 5482035b3afb..a59ad1c95556 100644 --- a/src/diffusers/pipelines/chroma/pipeline_chroma.py +++ b/src/diffusers/pipelines/chroma/pipeline_chroma.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -33,7 +32,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import ChromaPipelineOutput @@ -74,80 +73,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class ChromaPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/chroma/pipeline_chroma_img2img.py b/src/diffusers/pipelines/chroma/pipeline_chroma_img2img.py index 9afd4b9e1577..0dc81846f068 100644 --- a/src/diffusers/pipelines/chroma/pipeline_chroma_img2img.py +++ b/src/diffusers/pipelines/chroma/pipeline_chroma_img2img.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -33,7 +32,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import ChromaPipelineOutput @@ -72,94 +71,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class ChromaImg2ImgPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox.py b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox.py index 4ac33b24bbe1..23c78091a9cc 100644 --- a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox.py +++ b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox.py @@ -29,6 +29,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import CogVideoXPipelineOutput @@ -84,66 +85,6 @@ def get_resize_crop_region_for_grid(src, tgt_width, tgt_height): return (crop_top, crop_left), (crop_top + resize_height, crop_left + resize_width) -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin): r""" Pipeline for text-to-video generation using CogVideoX. diff --git a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_fun_control.py b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_fun_control.py index c1335839f848..fdbd17eaa25f 100644 --- a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_fun_control.py +++ b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_fun_control.py @@ -30,6 +30,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import CogVideoXPipelineOutput @@ -91,66 +92,6 @@ def get_resize_crop_region_for_grid(src, tgt_width, tgt_height): return (crop_top, crop_left), (crop_top + resize_height, crop_left + resize_width) -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class CogVideoXFunControlPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin): r""" Pipeline for controlled text-to-video generation using CogVideoX Fun. diff --git a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py index c523c9adec98..0f371f2ea795 100644 --- a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py +++ b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py @@ -31,6 +31,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_latents, retrieve_timesteps from .pipeline_output import CogVideoXPipelineOutput @@ -83,80 +84,6 @@ def get_resize_crop_region_for_grid(src, tgt_width, tgt_height): return (crop_top, crop_left), (crop_top + resize_height, crop_left + resize_width) -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class CogVideoXImageToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin): r""" Pipeline for image-to-video generation using CogVideoX. diff --git a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_video2video.py b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_video2video.py index 897dc6d1b70a..e48f973d6344 100644 --- a/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_video2video.py +++ b/src/diffusers/pipelines/cogvideo/pipeline_cogvideox_video2video.py @@ -30,6 +30,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_latents, retrieve_timesteps from .pipeline_output import CogVideoXPipelineOutput @@ -92,80 +93,6 @@ def get_resize_crop_region_for_grid(src, tgt_width, tgt_height): return (crop_top, crop_left), (crop_top + resize_height, crop_left + resize_width) -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin): r""" Pipeline for video-to-video generation using CogVideoX. diff --git a/src/diffusers/pipelines/cogview3/pipeline_cogview3plus.py b/src/diffusers/pipelines/cogview3/pipeline_cogview3plus.py index 304a5c5ad00b..d17ca5a149e5 100644 --- a/src/diffusers/pipelines/cogview3/pipeline_cogview3plus.py +++ b/src/diffusers/pipelines/cogview3/pipeline_cogview3plus.py @@ -26,6 +26,7 @@ from ...schedulers import CogVideoXDDIMScheduler, CogVideoXDPMScheduler from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import CogView3PipelineOutput @@ -55,66 +56,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class CogView3PlusPipeline(DiffusionPipeline): r""" Pipeline for text-to-image generation using CogView3Plus. diff --git a/src/diffusers/pipelines/consisid/pipeline_consisid.py b/src/diffusers/pipelines/consisid/pipeline_consisid.py index 3e6c149d7f80..59d3d36b7070 100644 --- a/src/diffusers/pipelines/consisid/pipeline_consisid.py +++ b/src/diffusers/pipelines/consisid/pipeline_consisid.py @@ -37,6 +37,8 @@ if is_opencv_available(): import cv2 +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + logger = logging.get_logger(__name__) # pylint: disable=invalid-name @@ -173,80 +175,6 @@ def get_resize_crop_region_for_grid(src, tgt_width, tgt_height): return (crop_top, crop_left), (crop_top + resize_height, crop_left + resize_width) -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class ConsisIDPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin): r""" Pipeline for image-to-video generation using ConsisID. diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet.py index fe0e69314cca..fa2e3addabf9 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet.py @@ -38,7 +38,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion.pipeline_output import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -99,66 +99,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionControlNetPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py index 12cc6f630d80..ed0f5b2950e0 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py @@ -37,7 +37,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from ..stable_diffusion import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -99,20 +99,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def prepare_image(image): if isinstance(image, torch.Tensor): # Batch single image diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py index 6de8e5747b02..d8cae6335d07 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py @@ -39,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from ..stable_diffusion import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -111,20 +111,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionControlNetInpaintPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint_sd_xl.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint_sd_xl.py index 397ab15715c2..02d3eb0e896b 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint_sd_xl.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint_sd_xl.py @@ -61,6 +61,7 @@ from ...utils import is_torch_xla_available +from ..pipeline_utils import retrieve_latents if is_torch_xla_available(): @@ -73,20 +74,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - EXAMPLE_DOC_STRING = """ Examples: ```py diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py index 8b37d38b9099..a7c341b445fe 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py @@ -63,6 +63,7 @@ from ...utils import is_torch_xla_available +from ..pipeline_utils import retrieve_timesteps if is_torch_xla_available(): @@ -121,66 +122,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLControlNetPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl_img2img.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl_img2img.py index 4d4845c5a0a3..b03d1192f006 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl_img2img.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl_img2img.py @@ -63,6 +63,7 @@ from ...utils import is_torch_xla_available +from ..pipeline_utils import retrieve_latents if is_torch_xla_available(): @@ -150,20 +151,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionXLControlNetImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_inpaint_sd_xl.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_inpaint_sd_xl.py index fb58b222112a..4d63cf356725 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_inpaint_sd_xl.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_inpaint_sd_xl.py @@ -57,7 +57,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from ..stable_diffusion_xl.pipeline_output import StableDiffusionXLPipelineOutput @@ -78,20 +78,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - EXAMPLE_DOC_STRING = """ Examples: ```py diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl.py index 5961d389effb..f4fc09269eb5 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl.py @@ -67,6 +67,7 @@ from ...utils import is_torch_xla_available +from ..pipeline_utils import retrieve_timesteps if is_torch_xla_available(): @@ -115,66 +116,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLControlNetUnionPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl_img2img.py b/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl_img2img.py index 8fedb6d8609a..e267d3d72871 100644 --- a/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl_img2img.py +++ b/src/diffusers/pipelines/controlnet/pipeline_controlnet_union_sd_xl_img2img.py @@ -59,7 +59,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from ..stable_diffusion_xl.pipeline_output import StableDiffusionXLPipelineOutput @@ -175,20 +175,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionXLControlNetUnionImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet.py b/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet.py index c763411ab5f7..90cbf1075333 100644 --- a/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet.py +++ b/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import torch @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..stable_diffusion_3.pipeline_output import StableDiffusion3PipelineOutput @@ -80,66 +79,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3ControlNetPipeline( DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, SD3IPAdapterMixin ): diff --git a/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet_inpainting.py b/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet_inpainting.py index c33cf979c6d8..8494b0a98d66 100644 --- a/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet_inpainting.py +++ b/src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet_inpainting.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import torch @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..stable_diffusion_3.pipeline_output import StableDiffusion3PipelineOutput @@ -101,66 +100,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3ControlNetInpaintingPipeline( DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, SD3IPAdapterMixin ): diff --git a/src/diffusers/pipelines/cosmos/pipeline_cosmos2_text2image.py b/src/diffusers/pipelines/cosmos/pipeline_cosmos2_text2image.py index 66490c2be159..ed3c91021ef0 100644 --- a/src/diffusers/pipelines/cosmos/pipeline_cosmos2_text2image.py +++ b/src/diffusers/pipelines/cosmos/pipeline_cosmos2_text2image.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Callable, Dict, List, Optional, Union import numpy as np @@ -40,6 +39,9 @@ def __init__(self, *args, **kwargs): ) +from ..pipeline_utils import retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -72,66 +74,6 @@ def __init__(self, *args, **kwargs): """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class Cosmos2TextToImagePipeline(DiffusionPipeline): r""" Pipeline for text-to-image generation using [Cosmos Predict2](https://github.com/nvidia-cosmos/cosmos-predict2). diff --git a/src/diffusers/pipelines/cosmos/pipeline_cosmos2_video2world.py b/src/diffusers/pipelines/cosmos/pipeline_cosmos2_video2world.py index 23a74ad00f93..5c35282fa599 100644 --- a/src/diffusers/pipelines/cosmos/pipeline_cosmos2_video2world.py +++ b/src/diffusers/pipelines/cosmos/pipeline_cosmos2_video2world.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Callable, Dict, List, Optional, Union import numpy as np @@ -41,6 +40,9 @@ def __init__(self, *args, **kwargs): ) +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -77,80 +79,6 @@ def __init__(self, *args, **kwargs): """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class Cosmos2VideoToWorldPipeline(DiffusionPipeline): r""" Pipeline for video-to-world generation using [Cosmos Predict2](https://github.com/nvidia-cosmos/cosmos-predict2). diff --git a/src/diffusers/pipelines/cosmos/pipeline_cosmos_text2world.py b/src/diffusers/pipelines/cosmos/pipeline_cosmos_text2world.py index f0aa1ecf0e0f..c7fef256701d 100644 --- a/src/diffusers/pipelines/cosmos/pipeline_cosmos_text2world.py +++ b/src/diffusers/pipelines/cosmos/pipeline_cosmos_text2world.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Callable, Dict, List, Optional, Union import numpy as np @@ -40,6 +39,9 @@ def __init__(self, *args, **kwargs): ) +from ..pipeline_utils import retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -69,66 +71,6 @@ def __init__(self, *args, **kwargs): """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class CosmosTextToWorldPipeline(DiffusionPipeline): r""" Pipeline for text-to-world generation using [Cosmos Predict1](https://github.com/nvidia-cosmos/cosmos-predict1). diff --git a/src/diffusers/pipelines/cosmos/pipeline_cosmos_video2world.py b/src/diffusers/pipelines/cosmos/pipeline_cosmos_video2world.py index cd5a734cc311..583c9bc7e701 100644 --- a/src/diffusers/pipelines/cosmos/pipeline_cosmos_video2world.py +++ b/src/diffusers/pipelines/cosmos/pipeline_cosmos_video2world.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Callable, Dict, List, Optional, Union import numpy as np @@ -41,6 +40,9 @@ def __init__(self, *args, **kwargs): ) +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -98,80 +100,6 @@ def __init__(self, *args, **kwargs): """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class CosmosVideoToWorldPipeline(DiffusionPipeline): r""" Pipeline for image-to-world and video-to-world generation using [Cosmos diff --git a/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion.py b/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion.py index 003e7482741f..b02526fa7045 100644 --- a/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion.py +++ b/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion.py @@ -39,7 +39,7 @@ unscale_lora_layers, ) from ....utils.torch_utils import randn_tensor -from ...pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ...pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .modeling_roberta_series import RobertaSeriesModelWithTransformation from .pipeline_output import AltDiffusionPipelineOutput @@ -90,66 +90,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AltDiffusionPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion_img2img.py b/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion_img2img.py index 64bd06d02ea7..2c9d83db2108 100644 --- a/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion_img2img.py +++ b/src/diffusers/pipelines/deprecated/alt_diffusion/pipeline_alt_diffusion_img2img.py @@ -42,7 +42,7 @@ unscale_lora_layers, ) from ....utils.torch_utils import randn_tensor -from ...pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ...pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .modeling_roberta_series import RobertaSeriesModelWithTransformation from .pipeline_output import AltDiffusionPipelineOutput @@ -80,20 +80,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.preprocess def preprocess(image): deprecation_message = "The preprocess method is deprecated and will be removed in diffusers 1.0.0. Please use VaeImageProcessor.preprocess(...) instead" @@ -118,66 +104,6 @@ def preprocess(image): return image -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class AltDiffusionImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_cycle_diffusion.py b/src/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_cycle_diffusion.py index 59c79e134e07..f898b1d710c4 100644 --- a/src/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_cycle_diffusion.py +++ b/src/diffusers/pipelines/deprecated/stable_diffusion_variants/pipeline_cycle_diffusion.py @@ -29,7 +29,7 @@ from ....schedulers import DDIMScheduler from ....utils import PIL_INTERPOLATION, USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers from ....utils.torch_utils import randn_tensor -from ...pipeline_utils import DiffusionPipeline +from ...pipeline_utils import DiffusionPipeline, retrieve_latents from ...stable_diffusion.pipeline_output import StableDiffusionPipelineOutput from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -61,20 +61,6 @@ def preprocess(image): return image -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def posterior_sample(scheduler, latents, timestep, clean_latents, generator, eta): # 1. get previous step value (=t-1) prev_timestep = timestep - scheduler.config.num_train_timesteps // scheduler.num_inference_steps diff --git a/src/diffusers/pipelines/easyanimate/pipeline_easyanimate.py b/src/diffusers/pipelines/easyanimate/pipeline_easyanimate.py index 92239c0d32f0..8bc249431273 100755 --- a/src/diffusers/pipelines/easyanimate/pipeline_easyanimate.py +++ b/src/diffusers/pipelines/easyanimate/pipeline_easyanimate.py @@ -31,6 +31,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import EasyAnimatePipelineOutput @@ -123,66 +124,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class EasyAnimatePipeline(DiffusionPipeline): r""" Pipeline for text-to-video generation using EasyAnimate. diff --git a/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_control.py b/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_control.py index f74a11f87d75..d42b62d1fca0 100755 --- a/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_control.py +++ b/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_control.py @@ -35,6 +35,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import EasyAnimatePipelineOutput @@ -225,66 +226,6 @@ def resize_mask(mask, latent, process_first_frame_only=True): return resized_mask -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class EasyAnimateControlPipeline(DiffusionPipeline): r""" Pipeline for text-to-video generation using EasyAnimate. diff --git a/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_inpaint.py b/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_inpaint.py index b16ef92d8e6b..6bee8e54da46 100755 --- a/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_inpaint.py +++ b/src/diffusers/pipelines/easyanimate/pipeline_easyanimate_inpaint.py @@ -35,6 +35,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps from .pipeline_output import EasyAnimatePipelineOutput @@ -267,66 +268,6 @@ def add_noise_to_reference_video(image, ratio=None, generator=None): return image -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class EasyAnimateInpaintPipeline(DiffusionPipeline): r""" Pipeline for text-to-video generation using EasyAnimate. diff --git a/src/diffusers/pipelines/flux/pipeline_flux.py b/src/diffusers/pipelines/flux/pipeline_flux.py index 5041e352f73d..25c06aa91f2e 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux.py +++ b/src/diffusers/pipelines/flux/pipeline_flux.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -71,79 +70,6 @@ """ -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/flux/pipeline_flux_control.py b/src/diffusers/pipelines/flux/pipeline_flux_control.py index 848d7bd39254..c2c67fe0b8e5 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_control.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_control.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -34,7 +33,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -83,80 +82,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxControlPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/flux/pipeline_flux_control_img2img.py b/src/diffusers/pipelines/flux/pipeline_flux_control_img2img.py index 262345c75afc..519c53b5e10c 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_control_img2img.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_control_img2img.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -33,7 +32,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -87,94 +86,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxControlImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin): r""" The Flux pipeline for image inpainting. diff --git a/src/diffusers/pipelines/flux/pipeline_flux_control_inpaint.py b/src/diffusers/pipelines/flux/pipeline_flux_control_inpaint.py index 6915a83a7ca7..1ab03f7ed985 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_control_inpaint.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_control_inpaint.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -43,7 +42,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -114,94 +113,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxControlInpaintPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/flux/pipeline_flux_controlnet.py b/src/diffusers/pipelines/flux/pipeline_flux_controlnet.py index 507ec687347c..20f7d2a30cbf 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_controlnet.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_controlnet.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import numpy as np @@ -41,7 +40,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -86,94 +85,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxControlNetPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin, FluxIPAdapterMixin): r""" The Flux pipeline for text-to-image generation. diff --git a/src/diffusers/pipelines/flux/pipeline_flux_controlnet_image_to_image.py b/src/diffusers/pipelines/flux/pipeline_flux_controlnet_image_to_image.py index 582c7bbad84e..a582501dcc98 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_controlnet_image_to_image.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_controlnet_image_to_image.py @@ -1,4 +1,3 @@ -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import numpy as np @@ -25,7 +24,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -81,94 +80,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxControlNetImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin): r""" The Flux controlnet pipeline for image-to-image generation. diff --git a/src/diffusers/pipelines/flux/pipeline_flux_controlnet_inpainting.py b/src/diffusers/pipelines/flux/pipeline_flux_controlnet_inpainting.py index f7f34ef231af..e96dd405dea4 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_controlnet_inpainting.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_controlnet_inpainting.py @@ -1,4 +1,3 @@ -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import numpy as np @@ -26,7 +25,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -83,94 +82,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxControlNetInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin): r""" The Flux controlnet pipeline for inpainting. diff --git a/src/diffusers/pipelines/flux/pipeline_flux_fill.py b/src/diffusers/pipelines/flux/pipeline_flux_fill.py index 5cb9c82204b2..e17ae602baea 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_fill.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_fill.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -34,7 +33,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -77,94 +76,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class FluxFillPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/flux/pipeline_flux_img2img.py b/src/diffusers/pipelines/flux/pipeline_flux_img2img.py index ab9140dae921..ab3f7a489027 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_img2img.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_img2img.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -41,7 +40,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -79,94 +78,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxImg2ImgPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixin, FluxIPAdapterMixin): r""" The Flux pipeline for image inpainting. diff --git a/src/diffusers/pipelines/flux/pipeline_flux_inpaint.py b/src/diffusers/pipelines/flux/pipeline_flux_inpaint.py index 3bfe82cf4382..f01499e89bb4 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_inpaint.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_inpaint.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -41,7 +40,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -75,94 +74,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FluxInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FluxIPAdapterMixin): r""" The Flux pipeline for image inpainting. diff --git a/src/diffusers/pipelines/flux/pipeline_flux_kontext.py b/src/diffusers/pipelines/flux/pipeline_flux_kontext.py index 94ae460afcd0..b4cda1041700 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_kontext.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_kontext.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -101,93 +100,6 @@ ] -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class FluxKontextPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/flux/pipeline_flux_kontext_inpaint.py b/src/diffusers/pipelines/flux/pipeline_flux_kontext_inpaint.py index b6f957981e14..6a15f865c7d2 100644 --- a/src/diffusers/pipelines/flux/pipeline_flux_kontext_inpaint.py +++ b/src/diffusers/pipelines/flux/pipeline_flux_kontext_inpaint.py @@ -1,7 +1,6 @@ # Copyright 2025 ZenAI. All rights reserved. # author: @vuongminh1907 -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -30,7 +29,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import FluxPipelineOutput @@ -125,93 +124,6 @@ ] -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class FluxKontextInpaintPipeline( DiffusionPipeline, FluxLoraLoaderMixin, diff --git a/src/diffusers/pipelines/hidream_image/pipeline_hidream_image.py b/src/diffusers/pipelines/hidream_image/pipeline_hidream_image.py index b6af23bca8fd..1e5cf3284f3e 100644 --- a/src/diffusers/pipelines/hidream_image/pipeline_hidream_image.py +++ b/src/diffusers/pipelines/hidream_image/pipeline_hidream_image.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect import math from typing import Any, Callable, Dict, List, Optional, Union @@ -32,7 +31,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler, UniPCMultistepScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import HiDreamImagePipelineOutput @@ -83,80 +82,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class HiDreamImagePipeline(DiffusionPipeline, HiDreamImageLoraLoaderMixin): model_cpu_offload_seq = "text_encoder->text_encoder_2->text_encoder_3->text_encoder_4->transformer->vae" _callback_tensor_inputs = ["latents", "prompt_embeds_t5", "prompt_embeds_llama3", "pooled_prompt_embeds"] diff --git a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_skyreels_image2video.py b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_skyreels_image2video.py index b50a6ae3ed27..6a42e0f78fff 100644 --- a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_skyreels_image2video.py +++ b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_skyreels_image2video.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import numpy as np @@ -27,7 +26,7 @@ from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents, retrieve_timesteps from .pipeline_output import HunyuanVideoPipelineOutput @@ -92,80 +91,6 @@ } -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class HunyuanSkyreelsImageToVideoPipeline(DiffusionPipeline, HunyuanVideoLoraLoaderMixin): r""" Pipeline for image-to-video generation using HunyuanVideo. diff --git a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py index 5c8e295eaf4c..39a2186bd7e9 100644 --- a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py +++ b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import numpy as np @@ -26,7 +25,7 @@ from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from .pipeline_output import HunyuanVideoPipelineOutput @@ -81,66 +80,6 @@ } -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class HunyuanVideoPipeline(DiffusionPipeline, HunyuanVideoLoraLoaderMixin): r""" Pipeline for text-to-video generation using HunyuanVideo. diff --git a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_framepack.py b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_framepack.py index 8006514f47ea..2aebbdc5ff7f 100644 --- a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_framepack.py +++ b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_framepack.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect import math from enum import Enum from typing import Any, Callable, Dict, List, Optional, Tuple, Union @@ -36,7 +35,7 @@ from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import HunyuanVideoFramepackPipelineOutput @@ -161,80 +160,6 @@ } -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class FramepackSamplingType(str, Enum): VANILLA = "vanilla" INVERTED_ANTI_DRIFTING = "inverted_anti_drifting" diff --git a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_image2video.py b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_image2video.py index aa04e6509730..10723472eef8 100644 --- a/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_image2video.py +++ b/src/diffusers/pipelines/hunyuan_video/pipeline_hunyuan_video_image2video.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Tuple, Union import numpy as np @@ -33,7 +32,7 @@ from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents, retrieve_timesteps from .pipeline_output import HunyuanVideoPipelineOutput @@ -144,80 +143,6 @@ def _expand_input_ids_with_image_tokens( } -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class HunyuanVideoImageToVideoPipeline(DiffusionPipeline, HunyuanVideoLoraLoaderMixin): r""" Pipeline for image-to-video generation using HunyuanVideo. diff --git a/src/diffusers/pipelines/kolors/pipeline_kolors.py b/src/diffusers/pipelines/kolors/pipeline_kolors.py index 948f73ed91eb..075f9091a7d7 100644 --- a/src/diffusers/pipelines/kolors/pipeline_kolors.py +++ b/src/diffusers/pipelines/kolors/pipeline_kolors.py @@ -25,7 +25,7 @@ from ...schedulers import KarrasDiffusionSchedulers from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from .pipeline_output import KolorsPipelineOutput from .text_encoder import ChatGLMModel from .tokenizer import ChatGLMTokenizer @@ -61,66 +61,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class KolorsPipeline(DiffusionPipeline, StableDiffusionMixin, StableDiffusionLoraLoaderMixin, IPAdapterMixin): r""" Pipeline for text-to-image generation using Kolors. diff --git a/src/diffusers/pipelines/kolors/pipeline_kolors_img2img.py b/src/diffusers/pipelines/kolors/pipeline_kolors_img2img.py index 67d49b9a8c5e..7375047bc862 100644 --- a/src/diffusers/pipelines/kolors/pipeline_kolors_img2img.py +++ b/src/diffusers/pipelines/kolors/pipeline_kolors_img2img.py @@ -26,7 +26,7 @@ from ...schedulers import KarrasDiffusionSchedulers from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import empty_device_cache, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from .pipeline_output import KolorsPipelineOutput from .text_encoder import ChatGLMModel from .tokenizer import ChatGLMTokenizer @@ -66,80 +66,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class KolorsImg2ImgPipeline(DiffusionPipeline, StableDiffusionMixin, StableDiffusionXLLoraLoaderMixin, IPAdapterMixin): r""" Pipeline for text-to-image generation using Kolors. diff --git a/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py b/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py index 1d7733982ef5..8d9b9fbb8e1a 100644 --- a/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py +++ b/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_img2img.py @@ -37,7 +37,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from ..stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker @@ -51,80 +51,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - EXAMPLE_DOC_STRING = """ Examples: ```py diff --git a/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_text2img.py b/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_text2img.py index 3e96b44663e1..a874d9fc4bbf 100644 --- a/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_text2img.py +++ b/src/diffusers/pipelines/latent_consistency_models/pipeline_latent_consistency_text2img.py @@ -36,7 +36,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion import StableDiffusionPipelineOutput, StableDiffusionSafetyChecker @@ -70,66 +70,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class LatentConsistencyModelPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/latte/pipeline_latte.py b/src/diffusers/pipelines/latte/pipeline_latte.py index 4d42a7049ec9..5004d89bfe2c 100644 --- a/src/diffusers/pipelines/latte/pipeline_latte.py +++ b/src/diffusers/pipelines/latte/pipeline_latte.py @@ -39,6 +39,7 @@ ) from ...utils.torch_utils import is_compiled_module, randn_tensor from ...video_processor import VideoProcessor +from ..pipeline_utils import retrieve_timesteps if is_torch_xla_available(): @@ -77,66 +78,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - @dataclass class LattePipelineOutput(BaseOutput): frames: torch.Tensor diff --git a/src/diffusers/pipelines/ltx/pipeline_ltx.py b/src/diffusers/pipelines/ltx/pipeline_ltx.py index bd23e657c408..e6780d77ab22 100644 --- a/src/diffusers/pipelines/ltx/pipeline_ltx.py +++ b/src/diffusers/pipelines/ltx/pipeline_ltx.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -27,7 +26,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import LTXPipelineOutput @@ -66,80 +65,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.rescale_noise_cfg def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): r""" diff --git a/src/diffusers/pipelines/ltx/pipeline_ltx_condition.py b/src/diffusers/pipelines/ltx/pipeline_ltx_condition.py index 537588f67c95..24074c29ed79 100644 --- a/src/diffusers/pipelines/ltx/pipeline_ltx_condition.py +++ b/src/diffusers/pipelines/ltx/pipeline_ltx_condition.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from dataclasses import dataclass from typing import Any, Callable, Dict, List, Optional, Tuple, Union @@ -29,7 +28,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents, retrieve_timesteps from .pipeline_output import LTXPipelineOutput @@ -134,94 +133,6 @@ def linear_quadratic_schedule(num_steps, threshold_noise=0.025, linear_steps=Non return torch.tensor(sigma_schedule[:-1]) -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.rescale_noise_cfg def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): r""" diff --git a/src/diffusers/pipelines/ltx/pipeline_ltx_image2video.py b/src/diffusers/pipelines/ltx/pipeline_ltx_image2video.py index 694378b4f040..3c7bed34a778 100644 --- a/src/diffusers/pipelines/ltx/pipeline_ltx_image2video.py +++ b/src/diffusers/pipelines/ltx/pipeline_ltx_image2video.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -28,7 +27,7 @@ from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import LTXPipelineOutput @@ -71,94 +70,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.rescale_noise_cfg def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): r""" diff --git a/src/diffusers/pipelines/ltx/pipeline_ltx_latent_upsample.py b/src/diffusers/pipelines/ltx/pipeline_ltx_latent_upsample.py index 1e94f6895f6e..b76afb4ec97f 100644 --- a/src/diffusers/pipelines/ltx/pipeline_ltx_latent_upsample.py +++ b/src/diffusers/pipelines/ltx/pipeline_ltx_latent_upsample.py @@ -21,7 +21,7 @@ from ...utils import deprecate, get_logger from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .modeling_latent_upsampler import LTXLatentUpsamplerModel from .pipeline_output import LTXPipelineOutput @@ -29,20 +29,6 @@ logger = get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class LTXLatentUpsamplePipeline(DiffusionPipeline): model_cpu_offload_seq = "" diff --git a/src/diffusers/pipelines/lucy/pipeline_lucy_edit.py b/src/diffusers/pipelines/lucy/pipeline_lucy_edit.py index 69f69d5768a8..0ab777515c6e 100644 --- a/src/diffusers/pipelines/lucy/pipeline_lucy_edit.py +++ b/src/diffusers/pipelines/lucy/pipeline_lucy_edit.py @@ -31,7 +31,7 @@ from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .pipeline_output import LucyPipelineOutput @@ -117,20 +117,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class LucyEditPipeline(DiffusionPipeline, WanLoraLoaderMixin): r""" Pipeline for video-to-video generation using Lucy Edit. diff --git a/src/diffusers/pipelines/lumina/pipeline_lumina.py b/src/diffusers/pipelines/lumina/pipeline_lumina.py index b59c265646cd..8d7e4893129d 100644 --- a/src/diffusers/pipelines/lumina/pipeline_lumina.py +++ b/src/diffusers/pipelines/lumina/pipeline_lumina.py @@ -38,7 +38,7 @@ replace_example_docstring, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_timesteps if is_torch_xla_available(): @@ -57,6 +57,7 @@ if is_ftfy_available(): import ftfy + EXAMPLE_DOC_STRING = """ Examples: ```py @@ -73,66 +74,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class LuminaPipeline(DiffusionPipeline): r""" Pipeline for text-to-image generation using Lumina-T2I. diff --git a/src/diffusers/pipelines/lumina2/pipeline_lumina2.py b/src/diffusers/pipelines/lumina2/pipeline_lumina2.py index 937803edbcbc..bc4d8804fcde 100644 --- a/src/diffusers/pipelines/lumina2/pipeline_lumina2.py +++ b/src/diffusers/pipelines/lumina2/pipeline_lumina2.py @@ -31,7 +31,7 @@ replace_example_docstring, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, calculate_shift, retrieve_timesteps if is_torch_xla_available(): @@ -60,80 +60,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class Lumina2Pipeline(DiffusionPipeline, Lumina2LoraLoaderMixin): r""" Pipeline for text-to-image generation using Lumina-T2I. diff --git a/src/diffusers/pipelines/mochi/pipeline_mochi.py b/src/diffusers/pipelines/mochi/pipeline_mochi.py index 5874a92c6f2f..6a6c35ba7f34 100644 --- a/src/diffusers/pipelines/mochi/pipeline_mochi.py +++ b/src/diffusers/pipelines/mochi/pipeline_mochi.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -26,7 +25,7 @@ from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from .pipeline_output import MochiPipelineOutput @@ -75,66 +74,6 @@ def linear_quadratic_schedule(num_steps, threshold_noise, linear_steps=None): return sigma_schedule -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class MochiPipeline(DiffusionPipeline, Mochi1LoraLoaderMixin): r""" The mochi pipeline for text-to-video generation. diff --git a/src/diffusers/pipelines/omnigen/pipeline_omnigen.py b/src/diffusers/pipelines/omnigen/pipeline_omnigen.py index 090cb46aace4..89bf475c8932 100644 --- a/src/diffusers/pipelines/omnigen/pipeline_omnigen.py +++ b/src/diffusers/pipelines/omnigen/pipeline_omnigen.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Callable, Dict, List, Optional, Union import numpy as np @@ -31,6 +30,9 @@ if is_torchvision_available(): from .processor_omnigen import OmniGenMultiModalProcessor +from ..pipeline_utils import retrieve_timesteps + + if is_torch_xla_available(): XLA_AVAILABLE = True else: @@ -56,66 +58,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class OmniGenPipeline( DiffusionPipeline, ): diff --git a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd.py b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd.py index de66871922e6..b626c9d84ec5 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd.py @@ -37,7 +37,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion.pipeline_output import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .pag_utils import PAGMixin @@ -102,66 +102,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionControlNetPAGPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_inpaint.py b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_inpaint.py index 3daaac328caa..085b7be67dea 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_inpaint.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_inpaint.py @@ -38,7 +38,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import empty_device_cache, is_compiled_module, randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from ..stable_diffusion import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .pag_utils import PAGMixin @@ -114,20 +114,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionControlNetPAGInpaintPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl.py b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl.py index 24cbab43c966..a614bc64011c 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl.py @@ -63,6 +63,7 @@ from ...utils import is_torch_xla_available +from ..pipeline_utils import retrieve_timesteps if is_torch_xla_available(): @@ -125,66 +126,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLControlNetPAGPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl_img2img.py b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl_img2img.py index a6df1b22c8b9..b2ca443209a0 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl_img2img.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd_xl_img2img.py @@ -63,6 +63,7 @@ from ...utils import is_torch_xla_available +from ..pipeline_utils import retrieve_latents if is_torch_xla_available(): @@ -151,20 +152,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionXLControlNetPAGImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_kolors.py b/src/diffusers/pipelines/pag/pipeline_pag_kolors.py index 1368358db6ba..137d7ac3712e 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_kolors.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_kolors.py @@ -28,7 +28,7 @@ from ..kolors.pipeline_output import KolorsPipelineOutput from ..kolors.text_encoder import ChatGLMModel from ..kolors.tokenizer import ChatGLMTokenizer -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from .pag_utils import PAGMixin @@ -66,66 +66,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class KolorsPAGPipeline( DiffusionPipeline, StableDiffusionMixin, StableDiffusionXLLoraLoaderMixin, IPAdapterMixin, PAGMixin ): diff --git a/src/diffusers/pipelines/pag/pipeline_pag_pixart_sigma.py b/src/diffusers/pipelines/pag/pipeline_pag_pixart_sigma.py index 9031877b5b8d..1bb0ad4bbdcc 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_pixart_sigma.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_pixart_sigma.py @@ -34,7 +34,7 @@ replace_example_docstring, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_timesteps from ..pixart_alpha.pipeline_pixart_alpha import ( ASPECT_RATIO_256_BIN, ASPECT_RATIO_512_BIN, @@ -81,66 +81,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class PixArtSigmaPAGPipeline(DiffusionPipeline, PAGMixin): r""" [PAG pipeline](https://huggingface.co/docs/diffusers/main/en/using-diffusers/pag) for text-to-image generation diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sana.py b/src/diffusers/pipelines/pag/pipeline_pag_sana.py index 9e91ccbe8006..52467e8ab748 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sana.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sana.py @@ -37,7 +37,7 @@ replace_example_docstring, ) from ...utils.torch_utils import get_device, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_timesteps from ..pixart_alpha.pipeline_pixart_alpha import ( ASPECT_RATIO_512_BIN, ASPECT_RATIO_1024_BIN, @@ -85,66 +85,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class SanaPAGPipeline(DiffusionPipeline, PAGMixin): r""" Pipeline for text-to-image generation using [Sana](https://huggingface.co/papers/2410.10629). This pipeline diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd.py b/src/diffusers/pipelines/pag/pipeline_pag_sd.py index 349d006aada3..b7801850be08 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd.py @@ -34,7 +34,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion.pipeline_output import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .pag_utils import PAGMixin @@ -94,66 +94,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionPAGPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_3.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_3.py index acb4e52340a6..ba9d7491569f 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_3.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_3.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import torch @@ -38,7 +37,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..stable_diffusion_3.pipeline_output import StableDiffusion3PipelineOutput from .pag_utils import PAGMixin @@ -73,66 +72,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3PAGPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, PAGMixin): r""" [PAG pipeline](https://huggingface.co/docs/diffusers/main/en/using-diffusers/pag) for text-to-image generation diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_3_img2img.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_3_img2img.py index e1819a79fb30..671aaae869a3 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_3_img2img.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_3_img2img.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import PIL.Image @@ -39,7 +38,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents, retrieve_timesteps from ..stable_diffusion_3.pipeline_output import StableDiffusion3PipelineOutput from .pag_utils import PAGMixin @@ -75,80 +74,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3PAGImg2ImgPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, PAGMixin): r""" [PAG pipeline](https://huggingface.co/docs/diffusers/main/en/using-diffusers/pag) for image-to-image generation diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_img2img.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_img2img.py index e9a846b5e2c4..d6d755befbb7 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_img2img.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_img2img.py @@ -37,7 +37,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from ..stable_diffusion.pipeline_output import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .pag_utils import PAGMixin @@ -75,80 +75,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionPAGImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_inpaint.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_inpaint.py index ee9d20f3682b..fe827c1ff0c1 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_inpaint.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_inpaint.py @@ -35,7 +35,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from ..stable_diffusion.pipeline_output import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .pag_utils import PAGMixin @@ -80,20 +80,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.rescale_noise_cfg def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): r""" @@ -121,66 +107,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionPAGInpaintPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_xl.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_xl.py index 6b62ddcc7ca5..71f92cacef81 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_xl.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_xl.py @@ -57,6 +57,9 @@ if is_invisible_watermark_available(): from ..stable_diffusion_xl.watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -113,66 +116,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLPAGPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_img2img.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_img2img.py index b6422b23648c..44eb3b36e0e3 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_img2img.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_img2img.py @@ -58,6 +58,9 @@ if is_invisible_watermark_available(): from ..stable_diffusion_xl.watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -117,80 +120,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLPAGImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_inpaint.py b/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_inpaint.py index 2e12a4a97fbe..1db23e04a7c7 100644 --- a/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_inpaint.py +++ b/src/diffusers/pipelines/pag/pipeline_pag_sd_xl_inpaint.py @@ -58,6 +58,9 @@ if is_invisible_watermark_available(): from ..stable_diffusion_xl.watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -130,80 +133,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLPAGInpaintPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py b/src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py index 61435b80ca5a..b7136dd88dde 100644 --- a/src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py +++ b/src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py @@ -25,7 +25,7 @@ from ...schedulers import DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler from ...utils import deprecate, is_torch_xla_available, logging from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin, retrieve_latents from ..stable_diffusion import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker from .image_encoder import PaintByExampleImageEncoder @@ -41,20 +41,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def prepare_mask_and_masked_image(image, mask): """ Prepares a pair (image, mask) to be consumed by the Paint by Example pipeline. This means that those inputs will be diff --git a/src/diffusers/pipelines/pipeline_utils.py b/src/diffusers/pipelines/pipeline_utils.py index 392d5fb3feb4..2749c71ab71c 100644 --- a/src/diffusers/pipelines/pipeline_utils.py +++ b/src/diffusers/pipelines/pipeline_utils.py @@ -2306,3 +2306,88 @@ def unfuse_qkv_projections(self, unet: bool = True, vae: bool = True): else: self.vae.unfuse_qkv_projections() self.fusing_vae = False + + +def retrieve_timesteps( + scheduler, + num_inference_steps: Optional[int] = None, + device: Optional[Union[str, torch.device]] = None, + timesteps: Optional[List[int]] = None, + sigmas: Optional[List[float]] = None, + **kwargs, +): + r""" + Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles + custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. + + Args: + scheduler (`SchedulerMixin`): + The scheduler to get timesteps from. + num_inference_steps (`int`): + The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` + must be `None`. + device (`str` or `torch.device`, *optional*): + The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. + timesteps (`List[int]`, *optional*): + Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, + `num_inference_steps` and `sigmas` must be `None`. + sigmas (`List[float]`, *optional*): + Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, + `num_inference_steps` and `timesteps` must be `None`. + + Returns: + `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the + second element is the number of inference steps. + """ + if timesteps is not None and sigmas is not None: + raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") + if timesteps is not None: + accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) + if not accepts_timesteps: + raise ValueError( + f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" + f" timestep schedules. Please check whether you are using the correct scheduler." + ) + scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) + timesteps = scheduler.timesteps + num_inference_steps = len(timesteps) + elif sigmas is not None: + accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) + if not accept_sigmas: + raise ValueError( + f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" + f" sigmas schedules. Please check whether you are using the correct scheduler." + ) + scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) + timesteps = scheduler.timesteps + num_inference_steps = len(timesteps) + else: + scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) + timesteps = scheduler.timesteps + return timesteps, num_inference_steps + + +def retrieve_latents( + encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" +): + if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": + return encoder_output.latent_dist.sample(generator) + elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": + return encoder_output.latent_dist.mode() + elif hasattr(encoder_output, "latents"): + return encoder_output.latents + else: + raise AttributeError("Could not access latents of provided encoder_output") + + +def calculate_shift( + image_seq_len, + base_seq_len: int = 256, + max_seq_len: int = 4096, + base_shift: float = 0.5, + max_shift: float = 1.16, +): + m = (max_shift - base_shift) / (max_seq_len - base_seq_len) + b = base_shift - m * base_seq_len + mu = image_seq_len * m + b + return mu diff --git a/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py b/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py index 1d718a4852a4..3b4962178918 100644 --- a/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py +++ b/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py @@ -34,7 +34,7 @@ replace_example_docstring, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_timesteps if is_torch_xla_available(): @@ -53,6 +53,7 @@ if is_ftfy_available(): import ftfy + EXAMPLE_DOC_STRING = """ Examples: ```py @@ -178,66 +179,6 @@ } -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class PixArtAlphaPipeline(DiffusionPipeline): r""" Pipeline for text-to-image generation using PixArt-Alpha. diff --git a/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_sigma.py b/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_sigma.py index bb169ac5c443..ca1832227f14 100644 --- a/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_sigma.py +++ b/src/diffusers/pipelines/pixart_alpha/pipeline_pixart_sigma.py @@ -34,7 +34,7 @@ replace_example_docstring, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_timesteps from .pipeline_pixart_alpha import ( ASPECT_RATIO_256_BIN, ASPECT_RATIO_512_BIN, @@ -122,66 +122,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class PixArtSigmaPipeline(DiffusionPipeline): r""" Pipeline for text-to-image generation using PixArt-Sigma. diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage.py index 33dc2039b986..9ede5706c89d 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -25,7 +24,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -56,79 +55,6 @@ """ -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class QwenImagePipeline(DiffusionPipeline, QwenImageLoraLoaderMixin): r""" The QwenImage pipeline for text-to-image generation. diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet.py index 5111096d93c1..03fa99d6ec84 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -26,7 +25,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -102,91 +101,6 @@ # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps class QwenImageControlNetPipeline(DiffusionPipeline, QwenImageLoraLoaderMixin): diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet_inpaint.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet_inpaint.py index 102a813ab582..e25e0c4687fe 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet_inpaint.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_controlnet_inpaint.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -26,7 +25,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -75,91 +74,6 @@ # Coped from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps class QwenImageControlNetInpaintPipeline(DiffusionPipeline, QwenImageLoraLoaderMixin): diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py index ed37b238c8c9..d89b08cc19b6 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect import math from typing import Any, Callable, Dict, List, Optional, Union @@ -26,7 +25,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -64,94 +63,6 @@ """ -# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def calculate_dimensions(target_area, ratio): width = math.sqrt(target_area * ratio) height = width / ratio diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_inpaint.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_inpaint.py index d54d1881fa4e..6bd28ac23c7b 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_inpaint.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_inpaint.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect import math from typing import Any, Callable, Dict, List, Optional, Union @@ -27,7 +26,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -65,94 +64,6 @@ """ -# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage_edit.calculate_dimensions def calculate_dimensions(target_area, ratio): width = math.sqrt(target_area * ratio) diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_plus.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_plus.py index ec203edf166c..515e04ceeb9a 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_plus.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit_plus.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect import math from typing import Any, Callable, Dict, List, Optional, Union @@ -26,7 +25,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -67,94 +66,6 @@ VAE_IMAGE_SIZE = 1024 * 1024 -# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def calculate_dimensions(target_area, ratio): width = math.sqrt(target_area * ratio) height = width / ratio diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_img2img.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_img2img.py index cb4c5d8016bb..6612f339b17c 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_img2img.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_img2img.py @@ -1,4 +1,3 @@ -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -11,7 +10,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -43,94 +42,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class QwenImageImg2ImgPipeline(DiffusionPipeline, QwenImageLoraLoaderMixin): r""" The QwenImage pipeline for text-to-image generation. diff --git a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_inpaint.py b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_inpaint.py index 1915c27eb2bb..9060c8681559 100644 --- a/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_inpaint.py +++ b/src/diffusers/pipelines/qwenimage/pipeline_qwenimage_inpaint.py @@ -1,4 +1,3 @@ -import inspect from typing import Any, Callable, Dict, List, Optional, Union import numpy as np @@ -12,7 +11,7 @@ from ...schedulers import FlowMatchEulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import QwenImagePipelineOutput @@ -46,94 +45,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.qwenimage.pipeline_qwenimage.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class QwenImageInpaintPipeline(DiffusionPipeline, QwenImageLoraLoaderMixin): r""" The QwenImage pipeline for text-to-image generation. diff --git a/src/diffusers/pipelines/sana/pipeline_sana.py b/src/diffusers/pipelines/sana/pipeline_sana.py index ac979305ca6d..75eca8f24075 100644 --- a/src/diffusers/pipelines/sana/pipeline_sana.py +++ b/src/diffusers/pipelines/sana/pipeline_sana.py @@ -40,7 +40,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import get_device, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..pixart_alpha.pipeline_pixart_alpha import ( ASPECT_RATIO_512_BIN, ASPECT_RATIO_1024_BIN, @@ -127,66 +127,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class SanaPipeline(DiffusionPipeline, SanaLoraLoaderMixin): r""" Pipeline for text-to-image generation using [Sana](https://huggingface.co/papers/2410.10629). diff --git a/src/diffusers/pipelines/sana/pipeline_sana_controlnet.py b/src/diffusers/pipelines/sana/pipeline_sana_controlnet.py index 55ed7b84ebdf..8163061f6270 100644 --- a/src/diffusers/pipelines/sana/pipeline_sana_controlnet.py +++ b/src/diffusers/pipelines/sana/pipeline_sana_controlnet.py @@ -40,7 +40,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import get_device, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..pixart_alpha.pipeline_pixart_alpha import ( ASPECT_RATIO_512_BIN, ASPECT_RATIO_1024_BIN, @@ -134,66 +134,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class SanaControlNetPipeline(DiffusionPipeline, SanaLoraLoaderMixin): r""" Pipeline for text-to-image generation using [Sana](https://huggingface.co/papers/2410.10629). diff --git a/src/diffusers/pipelines/sana/pipeline_sana_sprint.py b/src/diffusers/pipelines/sana/pipeline_sana_sprint.py index 62b978829271..876a8ae69049 100644 --- a/src/diffusers/pipelines/sana/pipeline_sana_sprint.py +++ b/src/diffusers/pipelines/sana/pipeline_sana_sprint.py @@ -40,7 +40,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import get_device, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..pixart_alpha.pipeline_pixart_alpha import ASPECT_RATIO_1024_BIN from .pipeline_output import SanaPipelineOutput @@ -78,66 +78,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class SanaSprintPipeline(DiffusionPipeline, SanaLoraLoaderMixin): r""" Pipeline for text-to-image generation using [SANA-Sprint](https://huggingface.co/papers/2503.09641). diff --git a/src/diffusers/pipelines/sana/pipeline_sana_sprint_img2img.py b/src/diffusers/pipelines/sana/pipeline_sana_sprint_img2img.py index 8899ed84c4e5..9654f3b40915 100644 --- a/src/diffusers/pipelines/sana/pipeline_sana_sprint_img2img.py +++ b/src/diffusers/pipelines/sana/pipeline_sana_sprint_img2img.py @@ -41,7 +41,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import get_device, is_torch_version, randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps from ..pixart_alpha.pipeline_pixart_alpha import ASPECT_RATIO_1024_BIN from .pipeline_output import SanaPipelineOutput @@ -61,6 +61,7 @@ if is_ftfy_available(): import ftfy + EXAMPLE_DOC_STRING = """ Examples: ```py @@ -84,66 +85,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class SanaSprintImg2ImgPipeline(DiffusionPipeline, SanaLoraLoaderMixin): r""" Pipeline for text-to-image generation using [SANA-Sprint](https://huggingface.co/papers/2503.09641). diff --git a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing.py b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing.py index d0a4e118ce43..2055c31b15d4 100644 --- a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing.py +++ b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing.py @@ -112,20 +112,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class SkyReelsV2DiffusionForcingPipeline(DiffusionPipeline, SkyReelsV2LoraLoaderMixin): """ Pipeline for Text-to-Video (t2v) generation using SkyReels-V2 with diffusion forcing. diff --git a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_i2v.py b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_i2v.py index 959cbb32f23a..ee81bee6aad8 100644 --- a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_i2v.py +++ b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_i2v.py @@ -32,7 +32,7 @@ from ...models import AutoencoderKLWan, SkyReelsV2Transformer3DModel from ...schedulers import UniPCMultistepScheduler from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .pipeline_output import SkyReelsV2PipelineOutput @@ -117,20 +117,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class SkyReelsV2DiffusionForcingImageToVideoPipeline(DiffusionPipeline, SkyReelsV2LoraLoaderMixin): """ Pipeline for Image-to-Video (i2v) generation using SkyReels-V2 with diffusion forcing. diff --git a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_v2v.py b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_v2v.py index 6fedfc795a40..2e138d4bff4e 100644 --- a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_v2v.py +++ b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_v2v.py @@ -13,7 +13,6 @@ # limitations under the License. import html -import inspect import math import re from copy import deepcopy @@ -31,7 +30,7 @@ from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .pipeline_output import SkyReelsV2PipelineOutput @@ -113,80 +112,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class SkyReelsV2DiffusionForcingVideoToVideoPipeline(DiffusionPipeline, SkyReelsV2LoraLoaderMixin): """ Pipeline for Video-to-Video (v2v) generation using SkyReels-V2 with diffusion forcing. diff --git a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_i2v.py b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_i2v.py index d59b4ce3cb17..13f8e60feeaa 100644 --- a/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_i2v.py +++ b/src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_i2v.py @@ -28,7 +28,7 @@ from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .pipeline_output import SkyReelsV2PipelineOutput @@ -110,20 +110,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class SkyReelsV2ImageToVideoPipeline(DiffusionPipeline, SkyReelsV2LoraLoaderMixin): r""" Pipeline for Image-to-Video (i2v) generation using SkyReels-V2. diff --git a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py index cb97f18efeff..9aff88f830d6 100644 --- a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py +++ b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py @@ -35,7 +35,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from .pipeline_output import StableDiffusionPipelineOutput from .safety_checker import StableDiffusionSafetyChecker @@ -92,65 +92,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py index e957c6661f87..5ada6d6040e5 100644 --- a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py +++ b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_depth2img.py @@ -38,7 +38,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, retrieve_latents if is_torch_xla_available(): @@ -51,20 +51,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.preprocess def preprocess(image): deprecation_message = "The preprocess method is deprecated and will be removed in diffusers 1.0.0. Please use VaeImageProcessor.preprocess(...) instead" diff --git a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py index 95d3ab06f02a..928d4ea3e815 100644 --- a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py +++ b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py @@ -39,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from . import StableDiffusionPipelineOutput from .safety_checker import StableDiffusionSafetyChecker @@ -83,19 +83,6 @@ """ -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def preprocess(image): deprecation_message = "The preprocess method is deprecated and will be removed in diffusers 1.0.0. Please use VaeImageProcessor.preprocess(...) instead" deprecate("preprocess", "1.0.0", deprecation_message, standard_warn=False) @@ -119,66 +106,6 @@ def preprocess(image): return image -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py index 148d7386a732..3eec67e44721 100644 --- a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py +++ b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py @@ -36,7 +36,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents, retrieve_timesteps from . import StableDiffusionPipelineOutput from .safety_checker import StableDiffusionSafetyChecker @@ -51,80 +51,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionInpaintPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py index 843d25d67c10..043523b77611 100644 --- a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py +++ b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py @@ -27,7 +27,7 @@ from ...schedulers import KarrasDiffusionSchedulers from ...utils import PIL_INTERPOLATION, deprecate, is_torch_xla_available, logging from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_latents from . import StableDiffusionPipelineOutput from .safety_checker import StableDiffusionSafetyChecker @@ -66,20 +66,6 @@ def preprocess(image): return image -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class StableDiffusionInstructPix2PixPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py index 66d5ffa6b849..8c28d4e1c567 100644 --- a/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py +++ b/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py @@ -27,7 +27,7 @@ from ...schedulers import EulerDiscreteScheduler from ...utils import deprecate, is_torch_xla_available, logging from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput, StableDiffusionMixin, retrieve_latents if is_torch_xla_available(): @@ -40,20 +40,6 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_upscale.preprocess def preprocess(image): warnings.warn( diff --git a/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3.py b/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3.py index 1618f89a49e3..c46e5a744ed1 100644 --- a/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3.py +++ b/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import torch @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_timesteps from .pipeline_output import StableDiffusion3PipelineOutput @@ -71,80 +70,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3Pipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, SD3IPAdapterMixin): r""" Args: diff --git a/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_img2img.py b/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_img2img.py index 7e97909f42ca..185c88aafa95 100644 --- a/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_img2img.py +++ b/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_img2img.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import PIL.Image @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import StableDiffusion3PipelineOutput @@ -77,94 +76,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3Img2ImgPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, SD3IPAdapterMixin): r""" Args: diff --git a/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_inpaint.py b/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_inpaint.py index bed596e57c34..0adae0385410 100644 --- a/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_inpaint.py +++ b/src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_inpaint.py @@ -12,7 +12,6 @@ # See the License for the specific language governing permissions and # limitations under the License. -import inspect from typing import Any, Callable, Dict, List, Optional, Union import torch @@ -40,7 +39,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .pipeline_output import StableDiffusion3PipelineOutput @@ -76,94 +75,6 @@ """ -# Copied from diffusers.pipelines.flux.pipeline_flux.calculate_shift -def calculate_shift( - image_seq_len, - base_seq_len: int = 256, - max_seq_len: int = 4096, - base_shift: float = 0.5, - max_shift: float = 1.15, -): - m = (max_shift - base_shift) / (max_seq_len - base_seq_len) - b = base_shift - m * base_seq_len - mu = image_seq_len * m + b - return mu - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusion3InpaintPipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingleFileMixin, SD3IPAdapterMixin): r""" Args: diff --git a/src/diffusers/pipelines/stable_diffusion_ldm3d/pipeline_stable_diffusion_ldm3d.py b/src/diffusers/pipelines/stable_diffusion_ldm3d/pipeline_stable_diffusion_ldm3d.py index c32121c88c9b..39884a678f11 100644 --- a/src/diffusers/pipelines/stable_diffusion_ldm3d/pipeline_stable_diffusion_ldm3d.py +++ b/src/diffusers/pipelines/stable_diffusion_ldm3d/pipeline_stable_diffusion_ldm3d.py @@ -37,7 +37,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -95,66 +95,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - @dataclass class LDM3DPipelineOutput(BaseOutput): """ diff --git a/src/diffusers/pipelines/stable_diffusion_panorama/pipeline_stable_diffusion_panorama.py b/src/diffusers/pipelines/stable_diffusion_panorama/pipeline_stable_diffusion_panorama.py index 295095947a12..ca03543c5b15 100644 --- a/src/diffusers/pipelines/stable_diffusion_panorama/pipeline_stable_diffusion_panorama.py +++ b/src/diffusers/pipelines/stable_diffusion_panorama/pipeline_stable_diffusion_panorama.py @@ -33,7 +33,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion import StableDiffusionPipelineOutput from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -95,66 +95,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionPanoramaPipeline( DeprecatedPipelineMixin, DiffusionPipeline, diff --git a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py index b97cf6f1f6f8..1d6c00aedf3d 100644 --- a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py +++ b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py @@ -58,6 +58,9 @@ if is_invisible_watermark_available(): from .watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -112,66 +115,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py index 44e8f4fe4b54..991a8b9e2a78 100644 --- a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py +++ b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py @@ -58,6 +58,9 @@ if is_invisible_watermark_available(): from .watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -115,80 +118,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLImg2ImgPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py index 18f8536a7510..a80f9a35ef95 100644 --- a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py +++ b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py @@ -59,6 +59,9 @@ if is_invisible_watermark_available(): from .watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_latents, retrieve_timesteps + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -142,80 +145,6 @@ def mask_pil_to_torch(mask, height, width): return mask -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLInpaintPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py index 58b008361782..06d9c0849cfa 100644 --- a/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py +++ b/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py @@ -46,6 +46,9 @@ if is_invisible_watermark_available(): from .watermark import StableDiffusionXLWatermarker +from ..pipeline_utils import retrieve_latents + + if is_torch_xla_available(): import torch_xla.core.xla_model as xm @@ -87,20 +90,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): """ Rescale `noise_cfg` according to `guidance_rescale`. Based on findings of [Common Diffusion Noise Schedules and diff --git a/src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py b/src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py index 6d9053faaec8..4ae35c07fac2 100644 --- a/src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py +++ b/src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py @@ -27,7 +27,7 @@ from ...utils import BaseOutput, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import is_compiled_module, randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_timesteps if is_torch_xla_available(): @@ -70,66 +70,6 @@ def _append_dims(x, target_dims): return x[(...,) + (None,) * dims_to_append] -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - @dataclass class StableVideoDiffusionPipelineOutput(BaseOutput): r""" diff --git a/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py b/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py index 1ce6987114a7..fd0229c9774e 100644 --- a/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py +++ b/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py @@ -38,7 +38,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker @@ -128,66 +128,6 @@ def _preprocess_adapter_image(image, height, width): return image -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionAdapterPipeline(DiffusionPipeline, StableDiffusionMixin, FromSingleFileMixin): r""" Pipeline for text-to-image generation using Stable Diffusion augmented with T2I-Adapter diff --git a/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py b/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py index 2802d690f3cc..b8df51b3722b 100644 --- a/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py +++ b/src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py @@ -50,7 +50,7 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DiffusionPipeline, StableDiffusionMixin, retrieve_timesteps from ..stable_diffusion_xl.pipeline_output import StableDiffusionXLPipelineOutput @@ -153,66 +153,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0): return noise_cfg -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - class StableDiffusionXLAdapterPipeline( DiffusionPipeline, StableDiffusionMixin, diff --git a/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py b/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py index 9d0b7e3dbc32..119da15ca9d8 100644 --- a/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py +++ b/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py @@ -34,7 +34,7 @@ ) from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin +from ..pipeline_utils import DeprecatedPipelineMixin, DiffusionPipeline, StableDiffusionMixin, retrieve_latents from . import TextToVideoSDPipelineOutput @@ -88,20 +88,6 @@ """ -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class VideoToVideoSDPipeline( DeprecatedPipelineMixin, DiffusionPipeline, diff --git a/src/diffusers/pipelines/visualcloze/pipeline_visualcloze_generation.py b/src/diffusers/pipelines/visualcloze/pipeline_visualcloze_generation.py index e12995106bcf..1c4461f56c68 100644 --- a/src/diffusers/pipelines/visualcloze/pipeline_visualcloze_generation.py +++ b/src/diffusers/pipelines/visualcloze/pipeline_visualcloze_generation.py @@ -32,9 +32,8 @@ unscale_lora_layers, ) from ...utils.torch_utils import randn_tensor -from ..flux.pipeline_flux_fill import calculate_shift, retrieve_latents, retrieve_timesteps from ..flux.pipeline_output import FluxPipelineOutput -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, calculate_shift, retrieve_latents, retrieve_timesteps from .visualcloze_utils import VisualClozeProcessor diff --git a/src/diffusers/pipelines/wan/pipeline_wan_i2v.py b/src/diffusers/pipelines/wan/pipeline_wan_i2v.py index b7fd0b05980f..cd0480b9935d 100644 --- a/src/diffusers/pipelines/wan/pipeline_wan_i2v.py +++ b/src/diffusers/pipelines/wan/pipeline_wan_i2v.py @@ -28,7 +28,7 @@ from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .pipeline_output import WanPipelineOutput @@ -44,6 +44,7 @@ if is_ftfy_available(): import ftfy + EXAMPLE_DOC_STRING = """ Examples: ```python @@ -110,20 +111,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin): r""" Pipeline for image-to-video generation using Wan. diff --git a/src/diffusers/pipelines/wan/pipeline_wan_vace.py b/src/diffusers/pipelines/wan/pipeline_wan_vace.py index 2b1890afec97..38f2ef335589 100644 --- a/src/diffusers/pipelines/wan/pipeline_wan_vace.py +++ b/src/diffusers/pipelines/wan/pipeline_wan_vace.py @@ -28,7 +28,7 @@ from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents from .pipeline_output import WanPipelineOutput @@ -124,20 +124,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class WanVACEPipeline(DiffusionPipeline, WanLoraLoaderMixin): r""" Pipeline for controllable generation using Wan. diff --git a/src/diffusers/pipelines/wan/pipeline_wan_video2video.py b/src/diffusers/pipelines/wan/pipeline_wan_video2video.py index 1a2d2e9c2232..4adbd71dac19 100644 --- a/src/diffusers/pipelines/wan/pipeline_wan_video2video.py +++ b/src/diffusers/pipelines/wan/pipeline_wan_video2video.py @@ -13,7 +13,6 @@ # limitations under the License. import html -import inspect from typing import Any, Callable, Dict, List, Optional, Union import regex as re @@ -28,7 +27,7 @@ from ...utils import is_ftfy_available, is_torch_xla_available, logging, replace_example_docstring from ...utils.torch_utils import randn_tensor from ...video_processor import VideoProcessor -from ..pipeline_utils import DiffusionPipeline +from ..pipeline_utils import DiffusionPipeline, retrieve_latents, retrieve_timesteps from .pipeline_output import WanPipelineOutput @@ -41,6 +40,7 @@ logger = logging.get_logger(__name__) # pylint: disable=invalid-name + if is_ftfy_available(): import ftfy @@ -97,80 +97,6 @@ def prompt_clean(text): return text -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps -def retrieve_timesteps( - scheduler, - num_inference_steps: Optional[int] = None, - device: Optional[Union[str, torch.device]] = None, - timesteps: Optional[List[int]] = None, - sigmas: Optional[List[float]] = None, - **kwargs, -): - r""" - Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles - custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`. - - Args: - scheduler (`SchedulerMixin`): - The scheduler to get timesteps from. - num_inference_steps (`int`): - The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps` - must be `None`. - device (`str` or `torch.device`, *optional*): - The device to which the timesteps should be moved to. If `None`, the timesteps are not moved. - timesteps (`List[int]`, *optional*): - Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed, - `num_inference_steps` and `sigmas` must be `None`. - sigmas (`List[float]`, *optional*): - Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed, - `num_inference_steps` and `timesteps` must be `None`. - - Returns: - `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the - second element is the number of inference steps. - """ - if timesteps is not None and sigmas is not None: - raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values") - if timesteps is not None: - accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accepts_timesteps: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" timestep schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - elif sigmas is not None: - accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys()) - if not accept_sigmas: - raise ValueError( - f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom" - f" sigmas schedules. Please check whether you are using the correct scheduler." - ) - scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs) - timesteps = scheduler.timesteps - num_inference_steps = len(timesteps) - else: - scheduler.set_timesteps(num_inference_steps, device=device, **kwargs) - timesteps = scheduler.timesteps - return timesteps, num_inference_steps - - -# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents -def retrieve_latents( - encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample" -): - if hasattr(encoder_output, "latent_dist") and sample_mode == "sample": - return encoder_output.latent_dist.sample(generator) - elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax": - return encoder_output.latent_dist.mode() - elif hasattr(encoder_output, "latents"): - return encoder_output.latents - else: - raise AttributeError("Could not access latents of provided encoder_output") - - class WanVideoToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin): r""" Pipeline for video-to-video generation using Wan.