faceswap

Why we included this project

FaceSwap is a long-running, complete open-source project for swapping faces in photos and video. The workflow is straightforward: extract faces from your source media, train a model on the two faces you want to exchange, then convert your footage with that model. A GUI sits on top of the command-line tools, so you can drive the whole process without writing scripts, and the documentation covers installation on Windows, Linux, and macOS, including GPU setup for both NVIDIA and AMD cards. Because the extract, train, and convert stages map directly onto the underlying machine-learning concepts, the codebase is also a practical way to learn how deep neural networks detect and synthesize faces. It is a desktop application you run locally rather than a hosted service, and the maintainers expect it to be used responsibly and with consent.