v1.0.2

This release is mainly a refactoring to publish a PyPI package so that parallax-maker is easier to use.

Changes

• created build scripts to package parallax-maker as wheel and publishing it to PyPI
• removed ZoeDepth as supported depth model since it is no longer maintained doesn't work with the new version of timm

v1.0.1

Features

• Added Docker support with complete Dockerfile and configuration
• Added Docker usage instructions to README.md

Fixes

• Updated webui.py to run server on all interfaces (0.0.0.0)
• Fixed Docker working directory and command path issues

Refactoring

• Updated export_gltf function to use Camera and ImageSlice objects
• Converted _dimension_at_depth method in ImageSlice class to staticmethod

Version 1.0.0 - Initial Release

First release of Parallax Maker.

Features

• Segmentation of images
  • Using depth models like Midas or ZeoDepth
  • Using instance segmentatio via Segment Anything with multiple positive and negative point selection
  • Adding and removing of cards, direct manipulation of depth values
• Inpainting
  • Inpainting of masks that can be padded and blurred
  • Replacing the masked regions with new images via image generation models like Stable Diffusion 1.0 XL, Stable Diffusion 3 Medium, Automatic1111 or ComyfUI endpoints as well as the StabilityAI API.
• 3D Export
  • Generation of glTF scenes that can be imported into Blender or Unreal Engine
  • Support for depth displacement of cards to generate more realistic 3D geometry
  • In browser 3D preview of the generated glTF scene.