FaceSwap

Why we included this project

This is a compact Python implementation of real-time 3D face swapping, originally written as a teaching exercise for a university course. Rather than leaning on a deep-learning black box, it walks through the classic pipeline: facial landmarks are detected, a 3D model with blendshapes is fitted using Gauss-Newton optimization, and the textured result is rendered with pygame and blended onto the live camera feed. The code is short enough to read in one sitting, which makes it a handy reference for anyone prototyping face-alignment or avatar features who wants to see how the pieces connect. One caveat: it is an educational project, not a maintained production library, so expect to download the dlib landmark model yourself and adapt the code before putting it to real use.