Open fkcptlst opened 1 month ago
I didn't do any wider training with this codebase.
you can see some things here https://github.com/johndpope/MegaPortrait-hack/issues/36
there's a newer repo emoportrait https://github.com/neeek2303/EMOPortraits the author who created the original code upgraded into this new repo
but the license is not favourable for commercial stuff.
you maybe interested in this work - https://github.com/johndpope/IMF
been focusing on stable diffusion + motion https://github.com/johndpope/DiPIR-hack (work in progress)
theres a lot of work happening in this space -
you can use this off the shelf model https://github.com/fudan-generative-vision/hallo
I built this repo to help recreate the VASA-1 paper - https://github.com/johndpope/vasa-1-hack
Hello, thanks for the great work! I wonder if there's any demo video for this implementation?
Additionally, have you evaluated the performance of this implementation? Does the digits match those reported in the paper?