#260 · Primary category: Video & Animation

MagicDance

behavior-generation cartoon-animation diffusion-models generative-ai generative-model image-editing video-editing video-generation

[ICML 2024] MagicPose(also known as MagicDance): Realistic Human Poses and Facial Expressions Retargeting with Identity-aware Diffusion

Project last updated:07/03/24

GitHub Stars

776

Forks

64

Contributors

2

License

Other

Why we included this project

MagicDance, published at ICML 2024 under the name MagicPose, retargets a person's pose and facial expressions onto a different identity while keeping the subject recognizable. It builds on Stable Diffusion with a ControlNet-style pose branch and an appearance-disentangled control module, which lets you animate a still photo from a pose sequence. The repo includes inference and training scripts, including multi-GPU training, and the README walks through preparing your own video data with OpenPose pose maps. It is research code, so you will need your own GPU setup and pose-detection pipeline, but for teams working on character animation or motion transfer it is a practical reference.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category