search

fudan-generative-vision / champ

Champ: Controllable and Consistent Human Image Animation with 3D Parametric Guidance
https://fudan-generative-vision.github.io/champ/
MIT License
3.91k stars 472 forks source link

my own test is bad! #45

Open min-star opened 5 months ago

min-star commented 5 months ago

[Uploading data.zip…]()

AricGamma commented 5 months ago

Thank you for your feedback. waiting for your test case.❤️❤️

Leoooo333 commented 5 months ago

Hi there, you can reopen your issue again when your get data prepared for sharing. Now your link only refer to this issue. It's recommended to use front-face reference image, and make your driving SMPLS matches the human's size as possible. Thanks for that and we are willing to see the failure cases!

min-star commented 5 months ago

data.zip

min-star commented 5 months ago

I use the example motion02，and the result is uploaded as data.zip,the face is collapse。

G-force78 commented 5 months ago

I use the example motion02，and the result is uploaded as data.zip,the face is collapse。

LOL..funny video, its because stable diffusion doesn't do faces very well unless its up close