Boese0601/MagicDance
A diffusion model that generates realistic human poses and facial expressions by retargeting from reference poses to target identities.
Not currently ranked — collecting fresh signals.
star history
MagicPose is an identity-aware diffusion model for realistic human pose and facial expression retargeting. It takes reference poses and transfers them to different character identities while maintaining natural motion and expressions. The model is published at ICML 2024 and includes implementation for pose-driven video generation.
Frequently asked
- What is Boese0601/MagicDance?
- A diffusion model that generates realistic human poses and facial expressions by retargeting from reference poses to target identities.
- Is MagicDance open source?
- Yes — Boese0601/MagicDance is an open-source project tracked on heatdrop.
- What language is MagicDance written in?
- Boese0601/MagicDance is primarily written in Python.
- How popular is MagicDance?
- Boese0601/MagicDance has 777 stars on GitHub.
- Where can I find MagicDance?
- Boese0601/MagicDance is on GitHub at https://github.com/Boese0601/MagicDance.