DiffusionAct: Controllable Diffusion Autoencoder for One-shot Face Reenactment

TL;DR

DiffusionAct introduces a diffusion autoencoder-based approach for high-quality, controllable one-shot face reenactment that preserves identity and appearance details without subject-specific fine-tuning.

Contribution

It leverages diffusion models to control facial pose and expressions in reenactment, outperforming GAN-based methods in quality and fidelity.

Findings

Achieves superior reenactment quality compared to GAN-based methods.

Supports one-shot, self, and cross-subject reenactment without fine-tuning.

Produces realistic images with preserved identity and appearance details.

Abstract

Video-driven neural face reenactment aims to synthesize realistic facial images that successfully preserve the identity and appearance of a source face, while transferring the target head pose and facial expressions. Existing GAN-based methods suffer from either distortions and visual artifacts or poor reconstruction quality, i.e., the background and several important appearance details, such as hair style/color, glasses and accessories, are not faithfully reconstructed. Recent advances in Diffusion Probabilistic Models (DPMs) enable the generation of high-quality realistic images. To this end, in this paper we present DiffusionAct, a novel method that leverages the photo-realistic image generation of diffusion models to perform neural face reenactment. Specifically, we propose to control the semantic space of a Diffusion Autoencoder (DiffAE), in order to edit the facial pose of the input images, defined as the head pose orientation and the facial expressions. Our method allows one-shot, self, and cross-subject reenactment, without requiring subject-specific fine-tuning. We compare against state-of-the-art GAN-, StyleGAN2-, and diffusion-based methods, showing better or on-par reenactment performance.