[Pipeline] Champ: Controllable and Consistent Human Image Animation with 3D Parametric Guidance

[a-r-r-o-w]

Model/Pipeline/Scheduler description

Opening this issue for sharing the news but it should already be widely covered across social media or will be soon. The authors compare performance against the mythical AnimateAnyone with 13k+ stars and 900 forks but no code, and show they are much superior in terms of generation quality. Very cool work and absolutely mind-blowing quality.

They already provide a diffusers-based implementation so there's nothing to add here I believe. Maybe one good idea would be to maintain a community README with list of great research repositories that utilize diffusers and provide ready to use pipelines for exploration. WDYT @yiyixuxu @sayakpaul?

Open source status

• [X] The model implementation is available.
• [X] The model weights are available (Only relevant if addition is not a scheduler).

Provide useful links for the implementation

Paper: https://arxiv.org/abs/2403.14781 Code: https://github.com/fudan-generative-vision/champ Project Page: https://fudan-generative-vision.github.io/champ/#/ Authors: @ShenhaoZhu Junming Leo Chen*, Zuozhuo Dai, Yinghui Xu, Xun Cao, Yao Yao, Hao Zhu, Siyu Zhu

[sayakpaul]

Thanks for notifying.

We have a way to fetch the repositories depending on diffusers. :-)

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[sayakpaul]

Closing because of inactivities.