Code and dataset for photorealistic Codec Avatars driven from audio - GitHub - facebookresearch/audio2photoreal: Code and dataset for photorealistic Codec Avatars driven from audio
They don’t have more articulated gestures down and the model doesn’t do a good job recognizing what part of a motion should be important in a gesture, which is funny because they said they’re better than before with pointing and hand waving and use an Italian style person in that example.
Everything I saw was talking slow and not articulating quickly.
They don’t have more articulated gestures down and the model doesn’t do a good job recognizing what part of a motion should be important in a gesture, which is funny because they said they’re better than before with pointing and hand waving and use an Italian style person in that example.
Everything I saw was talking slow and not articulating quickly.