Cross-identity reenactment with FLAME parameters extracted from EMOCA

[EvdoTheo]

Hello @ShenhanQian,

As i mentioned in #16 I have already modified EMOCA to capture webcam images to extract FLAME parameters. My goal is to input the expression and pose parameters on the Gaussian Avatars to animate an already-trained custom avatar. Do you think this is feasible, considering the necessary code modifications? if yes, which parts of the code should I adapt?

Thank you!