image2point.py

import torch
from PIL import Image
from point_e.diffusion.configs import diffusion_from_config, DIFFUSION_CONFIGS
from point_e.diffusion.sampler import PointCloudSampler
from point_e.models.configs import model_from_config, MODEL_CONFIGS
from point_e.models.download import load_checkpoint


def get_point_cloud_from_image(image_path):
    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

    print('creating base model...')
    base_name = 'base40M'  # use base300M or base1B for better results
    base_model = model_from_config(MODEL_CONFIGS[base_name], device)
    base_model.eval()
    base_diffusion = diffusion_from_config(DIFFUSION_CONFIGS[base_name])

    print('creating upsample model...')
    upsampler_model = model_from_config(MODEL_CONFIGS['upsample'], device)
    upsampler_model.eval()
    upsampler_diffusion = diffusion_from_config(DIFFUSION_CONFIGS['upsample'])

    print('downloading base checkpoint...')
    base_model.load_state_dict(load_checkpoint(base_name, device))

    print('downloading upsampler checkpoint...')
    upsampler_model.load_state_dict(load_checkpoint('upsample', device))

    print('creating sampler...')
    sampler = PointCloudSampler(
        device=device,
        models=[base_model, upsampler_model],
        diffusions=[base_diffusion, upsampler_diffusion],
        num_points=[1024, 4096 - 1024],
        aux_channels=['R', 'G', 'B'],
        guidance_scale=[3.0, 3.0],
    )

    img = Image.open(image_path)
    samples = None
    for x in (sampler.sample_batch_progressive(batch_size=1, model_kwargs=dict(images=[img]))):
        samples = x
    pc = sampler.output_to_point_clouds(samples)[0]
    return pc