MOST-GAN: 3D Morphable StyleGAN for Disentangled Face Image Manipulation

TL;DR

MOST-GAN introduces a novel framework combining 3D morphable models with style-based GANs to enable fully disentangled, photorealistic face image manipulation with explicit control over physical attributes like shape, pose, and lighting.

Contribution

It presents MOST-GAN, a framework that models physical face attributes explicitly, achieving disentangled, controllable, and photorealistic face image editing by design.

Findings

Enables extreme manipulation of lighting, expression, and pose.

Achieves photorealistic, disentangled face image editing.

Couples 3D morphable models with style-based GANs effectively.

Abstract

Recent advances in generative adversarial networks (GANs) have led to remarkable achievements in face image synthesis. While methods that use style-based GANs can generate strikingly photorealistic face images, it is often difficult to control the characteristics of the generated faces in a meaningful and disentangled way. Prior approaches aim to achieve such semantic control and disentanglement within the latent space of a previously trained GAN. In contrast, we propose a framework that a priori models physical attributes of the face such as 3D shape, albedo, pose, and lighting explicitly, thus providing disentanglement by design. Our method, MOST-GAN, integrates the expressive power and photorealism of style-based GANs with the physical disentanglement and flexibility of nonlinear 3D morphable models, which we couple with a state-of-the-art 2D hair manipulation network. MOST-GAN achieves photorealistic manipulation of portrait images with fully disentangled 3D control over their physical attributes, enabling extreme manipulation of lighting, facial expression, and pose variations up to full profile view.