Update 25.08.2024
-
Removed face parsing
-
Choose driving or target audio
-
Re-encode driving video to match target frame rate to keep driving audio in sync with target animation
-
Still work in progress
Another extended tpsmm version for Video-2-Video reenactment/portrait animation.
Maybe outdated models... Work in progress, some things to be added for better results.
Use driving video frontal face with not too much head movement for best results
Some results: (stock videos from www.pexels.com)