There were no virtual environments used, I will need to recode a version that does use venv environments. For cleaner code and less issues with it reinstalling packages or downgrading/upgrading packages you need for another project. I will make sure to implement this if I continue working on this project some day.
DeepFakeGui is a gui for the first order motion model it uses ffmpeg to prepare images automatically, and adds the audio back to the video, using ffmpeg.
make sure you have a dataset from the original project from Google drive or Yandex, i recommend vox-adv-cpk, which is for facial animation (from original repo)
FOM has some issues on newer version of python, this gui aims to fix that by patching the files with simple changes to make it compatible. there will be warnings, but errors should be gone from the training.
some things are not automatically installed, you will need to install them yourself
py -3.8 -m pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
f1
to open the output folder automaticallyvox-adv-cpk.pth.tar
you should put vox-adv
f1
)