3D portraits from 2D images
Researchers at the universities of Nottingham and Kingston in the UK have developed an AI tool that turns 2D portraits into 3D models. Unlike other, similar tools, this one can produce a good likeness based on a single photo. It uses a convolutional neural network (CNN) fueled by a library of 2D pictures of faces associated to their 3D models. It also works with video images, as seen in the example below. We’re thinking of all sorts of potential applications for this program, for example to create 3D avatars for video games. You can test it here, and the code is available on GitHub.
⇨ University of Nottingham, Aaron S. Jackson, “Large pose 3D face reconstruction from a single image via direct volumetric CNN regression.”