search

OpenTalker / SadTalker

[CVPR 2023] SadTalker：Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation
https://sadtalker.github.io/
Other
11.96k stars 2.22k forks source link

Request for an extention for oobabooga webui plz #248

Open sarutobiumon opened 1 year ago

sarutobiumon commented 1 year ago

https://github.com/oobabooga/text-generation-webui/discussions/1449

vinthony commented 1 year ago

very cool! I will work on this with a faster implementation recently.

johndpope commented 1 year ago

not sure anyone noticed - but the model seems to struggle to smile / laugh outlandishly. was the updated pytorch model used for this repo? How can we help retrain model to be more expressive?

Screenshot 2023-04-27 at 6 37 24 pm
vinthony commented 1 year ago

We use the PyTorch version, which seems to be the SOTA model of the open-sourced solution.

sarutobiumon commented 1 year ago

very cool! I will work on this with a faster implementation recently.

thank you sir!