Face Orbitals: Adapting Deep Learning models to Generate Artificial Faces
Faces are an important stimulus for many experiments in psychology, and new technologies let us create artificial faces that are highly realistic. We have modified StyleGAN—a deep learning algorithm that creates face pictures, so that it will produce a smooth “circle” of computer-generated faces. These are randomly generated from a 512-dimension “face space,” allowing for an incredible diversity of face types. Scalable parameters allow precise control over how smoothly the faces change around the circle.