Labels follow 24 parts of the face, using the landmarks in video pixels with an orthographic camera. A bit of 2D physics keeps the labels apart and lets them trail the face.
Loading...