search

Zielon / INSTA

INSTA - Instant Volumetric Head Avatars [CVPR2023]
https://zielon.github.io/insta/
Other
436 stars 37 forks source link

Reenactment failure: the number of faces (8090) of provided canonical.obj is not equal to that of mesh (9766) from Metrical Photometric Tracker #28

Closed tsuJohc closed 12 months ago

tsuJohc commented 12 months ago

Hi, thanks for sharing the excellent work!

I want to reenact the provided avatar with new expressions. So I need to use Metrical Photometric Tracker to track FLAME parameters of expression a from a video of another avatar and generate the mesh with the combination of canonical shape parameters and expression parameters of expression a.

However, the strange thing is that the number of faces(8090) of canonical mesh provided is not equal to that (9766) of output mesh from Metrical Photometric Tracker. I also found that the number of faces of output mesh from DECA is also 9766. Do you have any ideas on this problem?

Besides, could you please provide the FLAME parameters of each frame and canonical shape parameter in your dataset, which will make the reenactment easier (if so, no FLAME tracker is needed)?

Zielon commented 12 months ago

True, sorry for the confusion. INSTA uses a different number of faces. You can use this script to generate correct input geometry: https://github.com/Zielon/INSTA/blob/master/scripts/simplification.py.

For more information, you can take a look here https://github.com/Zielon/INSTA#dataset-generation.