Notice: This Repo is now in Read Only mode. Information here is likely to be increasingly outdated. Discussion about Faceswap usage should be redirected to our Forums at https://www.faceswap.dev
This repo has been opened for users playing with the project.
Sadly the project is not yet ready to use for non technical people, it will improve over time, so check for updates regularly and discuss with other.
- Discuss here about your experience of the project
- Help people having trouble running it
- Add guidelines and help for newcomers
If you are a more experienced user you can help by proposing solutions for a better user-experience. Ideally, the goal is to get to a release that can be used to non technical people.
Get the code from the main 'faceswap' repo and set it up (setup section below)
The project has multiple entry points. You will have to:
- Gather photos (or use the one provided in the training data provided below)
- Extract faces from your raw photos
- Train a model on your photos (or use the one provided in the training data provided below)
- Convert your sources with the mode
Read the full instructions at https://github.com/deepfakes/faceswap/blob/master/USAGE.md
Run python faceswap.py extract -h.
Run python faceswap.py train -h.
Run python faceswap.py convert -h.
Follow the directions at https://github.com/deepfakes/faceswap/blob/master/INSTALL.md
Main Requirements: Python 3.6 Opencv Tensorflow Keras
You also need an Nvidia GPU with CUDA support for best performance. Minimum is 4gb of Vram.