SimSwap

Why we included this project

SimSwap is the official PyTorch release behind the ACM Multimedia paper of the same name. One trained model swaps an arbitrary source face onto images or video while keeping the target's expression and gaze, and you do not need to fine-tune it for each new identity. That generality is why it keeps appearing in media tooling, creative software, and research projects. The repo ships training and inference code, a higher-resolution SimSwap-HQ variant, and Colab notebooks, so you can try it before deeper integration. Worth noting: the code is under a non-commercial Creative Commons license and is explicitly for technical and academic use, which matters if a commercial product is the end goal.