HDRFace: Rethinking Face Restoration with High-Dimensional Representation

2026-05-14

AI Summary

This work addresses the challenge of face restoration under severe degradation, where critical identity-preserving details are often lost due to extreme information deficiency. To tackle this, we propose HDRFace, a framework that enhances conditional generation through high-dimensional semantic representations. Specifically, an off-the-shelf restorer first produces an intermediate result, and a pretrained high-dimensional feature encoder extracts fine-grained representations from both the degraded input and the intermediate output, which are then injected as additional conditions into the generative process. Without altering the backbone architecture, our method introduces a novel Structure-and-Details-aware Fusion Mechanism (SDFM) that adaptively strengthens global structural constraints during coarse modeling and representation-guided detail synthesis in the refinement stage. Extensive experiments demonstrate that HDRFace consistently improves both restoration quality and identity fidelity across diverse generative backbones, including Stable Diffusion V2.1-base and Qwen-Image.

Abstract

Face restoration under complex degradations still remains an ill-posed inverse problem due to severe information loss. Although diffusion models benefit from strong generative priors, most methods still condition only on low-quality inputs, making it difficult to recover identity-critical details under heavy degradations. In this work, we propose HDRFace, a High-Dimensional Representation conditioned Face restoration framework that injects semantically rich priors into the conditional flow without modifying the generative backbone. Our pipeline first obtains a structurally reliable intermediate restoration with an off-the-shelf restorer, then uses a pretrained high-dimensional feature encoder to extract fine-grained facial representations from both the low-quality input and the intermediate result, and injects them as additional conditions for generation. We further introduce SDFM, a Structure-Detail aware adaptive Fusion Mechanism that emphasizes global constraints during structure modeling and strengthens representation guidance during detail synthesis, balancing structural consistency and detail fidelity. To validate the generalization ability of our method, we implement the proposed framework on two generative models, SD V2.1-base and Qwen-Image, and consistently observe stable and coherent performance gains across different architectures.

Problem

• face restoration
• complex degradations
• ill-posed inverse problem
• identity-critical details
• information loss

Innovation

• High-Dimensional Representation
• Face Restoration
• Conditional Diffusion
• Structure-Detail Fusion
• Generative Prior