Skip to content

NOTICE.md

Latest commit

 

History

History
183 lines (128 loc) · 6.54 KB

NOTICE.md

File metadata and controls

183 lines (128 loc) · 6.54 KB
Raw

NOTICE — ComfyUI-WanAnimatePreprocessV2

Copyright (c) 2025-2026 Code2Collapse (https://github.com/Code2Collapse)
License: Apache License 2.0 (see LICENSE)

Project author: Code2Collapse

This project is a derivative work based on:

  1. kijai's ComfyUI-WanAnimatePreprocess — original YOLO+ViTPose+Wan pipeline
    Author: kijai (Jukka Seppänen) — https://github.com/kijai
    Repository: https://github.com/kijai/ComfyUI-WanAnimatePreprocess
    License: Apache 2.0

  2. steven850's improved nodes — added CLAHE preprocessing, Gaussian blur,
    temporal face-bbox smoothing, constant-size face crop, detection threshold
    controls. Submitted as an attachment to issue #10 of kijai's repo.
    Author: steven850 — https://github.com/steven850
    Issue: kijai/ComfyUI-WanAnimatePreprocess#10
    License: Apache 2.0 (contribution to an Apache-2.0 repository)

Additions by Code2Collapse on top of the kijai/steven850 base:

  • Iris/pupil detection (gradient voting, Timm-Barth 2011 inspired multi-strategy fallback)
  • MediaPipe FaceMesh 478-point pipeline with iris/gaze/lip tracking
  • Protobuf ≥5.x compatibility fix for mediapipe ≤0.10.x
  • Lip openness ratio output
  • Renamed to V2 namespace with additional RETURN_TYPES

This project includes code and data derived from the following third-party works. Their copyrights and licenses are listed below.

1. Wan — Alibaba Wan Team

Repository owners: The Alibaba Wan Team Authors
Contact: https://github.com/Wan-Video

The following files are directly derived from the Wan open-source project and carry their original copyright notices intact (as required by Apache 2.0):

  • pose_utils/human_visualization.py — Copyright 2024-2025 The Alibaba Wan Team Authors.
  • pose_utils/pose2d_utils.py — Copyright 2024-2025 The Alibaba Wan Team Authors.
  • models/onnx_models.py — Copyright 2024-2025 The Alibaba Wan Team Authors.
  • retarget_pose.py — Copyright 2024-2025 The Alibaba Wan Team Authors.
  • utils.py — Copyright 2024-2025 The Alibaba Wan Team Authors.

Repository: https://github.com/Wan-Video/Wan2.1
License: Apache License 2.0
https://github.com/Wan-Video/Wan2.1/blob/main/LICENSE

2. kijai — ComfyUI-WanAnimatePreprocess

Author: kijai (Jukka Seppänen) — https://github.com/kijai
Repository: https://github.com/kijai/ComfyUI-WanAnimatePreprocess
License: Apache License 2.0

nodes.py and models/onnx_models.py in this pack are derivative works based on kijai's original ComfyUI node wrappers for the Wan Animate preprocessing pipeline. kijai's code wraps the Alibaba Wan Team's preprocess logic as ComfyUI nodes.

3. steven850 — Improved pose and face detection

Author: steven850 — https://github.com/steven850
Source: Issue #10 of kijai/ComfyUI-WanAnimatePreprocess
kijai/ComfyUI-WanAnimatePreprocess#10
License: Apache License 2.0 (contribution to Apache-2.0 repository)

The following improvements in nodes.py originate from steven850's contribution posted as issue #10:

  • CLAHE contrast enhancement preprocessing
  • Gaussian blur preprocessing for YOLO/ViTPose stability
  • Motion-adaptive temporal smoothing for face bounding boxes
  • Constant-size face crop with center-tracking
  • Detection threshold and pose threshold parameters
  • Core 8-stage pipeline architecture (YOLO → ViTPose → pose metas → face bbox → smooth → crop → package)

4. One-to-All-Animation (CVPR 2026)

Repository owners / Authors:
Shijun Shi (Jiangnan University), Jing Xu (USTC), Zhihang Li (CAS),
Chunli Peng (BUPT), Xiaoda Yang (Zhejiang University), Lijing Lu (CAS),
Kai Hu (Jiangnan University), Jiangning Zhang (Zhejiang University)
Contact: ssj180123@gmail.com
Repository: https://github.com/ssj9596/One-to-All-Animation

The following files are adapted from this project:

  • onetoall/infer_function.py — pose format conversion utilities
  • onetoall/utils.py — pose retargeting and scale-and-translate logic

License: Apache License 2.0
https://github.com/ssj9596/One-to-All-Animation/blob/main/LICENSE

Modifications by Code2Collapse (2025-2026):

  • Adapted for ComfyUI-WanAnimatePreprocessV2 node integration
  • Added aaposemeta_obj_to_dwpose for AAPoseMeta object support

Citation (if used in research):

@article{shi2025one,
  title={One-to-All Animation: Alignment-Free Character Animation and Image Pose Transfer},
  author={Shi, Shijun and Xu, Jing and Li, Zhihang and Peng, Chunli and Yang, Xiaoda and
          Lu, Lijing and Hu, Kai and Zhang, Jiangning},
  journal={arXiv preprint arXiv:2511.22940},
  year={2025}
}

5. MediaPipe (Google LLC)

Used for 478-point FaceMesh landmark detection in nodes.py.

Repository: https://github.com/google-ai-edge/mediapipe
Copyright: Copyright 2023 The MediaPipe Authors
License: Apache License 2.0
https://github.com/google-ai-edge/mediapipe/blob/master/LICENSE

6. ViTPose (MMLAB / OpenMMLab)

ONNX models for 2D human pose estimation are derived from the ViTPose project.

Repository: https://github.com/ViTAE-Transformer/ViTPose
Copyright: Copyright (c) 2022 ViTAE-Transformer
License: Apache License 2.0
https://github.com/ViTAE-Transformer/ViTPose/blob/main/LICENSE

7. ONNX Runtime (Microsoft)

Used as the ONNX inference backend for ViTPose and YOLO models.

Repository: https://github.com/microsoft/onnxruntime
Copyright: Copyright (c) Microsoft Corporation
License: MIT License
https://github.com/microsoft/onnxruntime/blob/main/LICENSE

8. DWPose / ControlNet Auxiliary Preprocessors

The pose data structures in onetoall/infer_function.py use the DWPose format developed by IDEA Research / ControlNet contributors.

Repository: https://github.com/IDEA-Research/DWPose
Copyright: Copyright (c) 2023 IDEA Research
License: Apache License 2.0

9. OpenCV

Repository: https://github.com/opencv/opencv
License: Apache License 2.0 (since OpenCV 4.5)

10. PyTorch

Repository: https://github.com/pytorch/pytorch
Copyright: Copyright (c) 2016-2024 Facebook, Inc. and its affiliates
License: BSD-style license — https://github.com/pytorch/pytorch/blob/main/LICENSE

Model Weights Notice

ViTPose and YOLO ONNX model weights are third-party assets. Users are responsible for reviewing the license of each model checkpoint before commercial use.