HeadGAN: One-shot Neural Head Synthesis and Editing
This work is significant for researchers and practitioners in computer graphics and animation seeking to generate realistic and controllable human head animations from limited input, offering improvements over existing methods.
The paper addresses the challenge of head reenactment from a single reference image, aiming to improve photo-realism, identity preservation, and pose/expression transfer. HeadGAN, their proposed system, conditions synthesis on 3D face representations and incorporates audio features for enhanced mouth movements.
Recent attempts to solve the problem of head reenactment using a single reference image have shown promising results. However, most of them either perform poorly in terms of photo-realism, or fail to meet the identity preservation problem, or do not fully transfer the driving pose and expression. We propose HeadGAN, a novel system that conditions synthesis on 3D face representations, which can be extracted from any driving video and adapted to the facial geometry of any reference image, disentangling identity from expression. We further improve mouth movements, by utilising audio features as a complementary input. The 3D face representation enables HeadGAN to be further used as an efficient method for compression and reconstruction and a tool for expression and pose editing.