DVFace: Spatio-Temporal Dual-Prior Diffusion for Video Face Restoration

TL;DR

DVFace is a novel one-step diffusion framework that leverages spatio-temporal priors for high-quality, temporally consistent video face restoration, outperforming existing methods.

Contribution

The paper introduces a dual-codebook design and asymmetric fusion module enabling effective one-step diffusion for video face restoration.

Findings

DVFace achieves superior restoration quality on benchmarks.

It maintains better temporal consistency than prior methods.

It preserves facial identity effectively during restoration.

Abstract

Video face restoration aims to enhance degraded face videos into high-quality results with realistic facial details, stable identity, and temporal coherence. Recent diffusion-based methods have brought strong generative priors to restoration and enabled more realistic detail synthesis. However, existing approaches for face videos still rely heavily on generic diffusion priors and multi-step sampling, which limit both facial adaptation and inference efficiency. These limitations motivate the use of one-step diffusion for video face restoration, yet achieving faithful facial recovery alongside temporally stable outputs remains challenging. In this paper, we propose, DVFace, a one-step diffusion framework for real-world video face restoration. Specifically, we introduce a spatio-temporal dual-codebook design to extract complementary spatial and temporal facial priors from degraded videos. We further propose an asymmetric spatio-temporal fusion module to inject these priors into the diffusion backbone according to their distinct roles. Evaluation on various benchmarks shows that DVFace delivers superior restoration quality, temporal consistency, and identity preservation compared to recent methods. Code: https://github.com/zhengchen1999/DVFace.