ECCV2020

We propose a new method to reconstruct a mesh-based avatar from RGB-D video taking albedo and lighting into account.