yanderifier

Why we included this project

This project wraps the first-order motion model in a double-click desktop app. You give it a still image of a face and a video clip, and it animates the face to follow the movement in the video, with no command line and no manual install of the deep learning stack. The bundled environment sets up the NVIDIA CUDA or CPU path for you, which clears the biggest hurdle most newcomers hit with this kind of work. It fits classroom demos of neural face reenactment well, and hobbyists who never managed to get the original research code running finally have something that works out of the box. One honest caveat: development has stalled and the maintainers are actively looking for help, so treat it as a finished tool for experimentation, not a base for new production work.