search

postech-ami / CLIP-Actor

[ECCV'22] Official PyTorch Implementation of "CLIP-Actor: Text-Driven Recommendation and Stylization for Animating Human Meshes"
https://clip-actor.github.io
MIT License
108 stars 5 forks source link

How to make a A-pose model for comparison #5

Closed zeng-yifei closed 1 year ago

zeng-yifei commented 1 year ago

I have tried to using a prompt like "Obama is standing", but can't produce a reasonable standing pose obj like A-pose.

How could I create a model that only stands and doing nothing?

Youwang-Kim commented 1 year ago

Hi, thank you for the interest in our project.

Regarding the A-pose (or T-pose), you could just dump the retrieved motion with zeroes: https://github.com/postech-ami/CLIP-Actor/blob/c1b3f1a0a6d5ea75f03ee58807ff0abfd965ab13/clip_actor.py#L415

For example, you can do it like below

body_pose, right_hand_pose, left_hand_pose = torch.zeros_like(body_pose, device=device), torch.zeros_like(right_hand_pose, device=device), torch.zeros_like(left_hand_pose, device=device)

Hope this helps.

Youwang-Kim commented 1 year ago

Please reopen this issue if you need more help regarding this.

zeng-yifei commented 1 year ago

I didn't expect such a quick reply. :-)

Thank you very much! It works finally!