In this project, our goal is to reconstruct the detailed 3D face geometry. The input can be a single face image or a talking face video. In the case of video, we can leverage the temporal information from multiple frames, as well as the audio signal.