Yufeng Zheng

Results 3 comments of Yufeng Zheng

I'm not the author but I think the model is expecting a single cropped face image, not a whole body or multiple faces.

Yes. IMavatar's output is aligned with FLAME except for a scaling factor.

1. I was assuming that the output is the reconstructed head SDF, and the input is the training video. 2. You can either sample from the mesh, or directly use...