PixRestore: Unified Image Restoration via Pixel Diffusion Transformer

2026-08-17

AI Summary

This study addresses the issues of detail loss and generation inconsistency in latent diffusion models for unified image inpainting by proposing a VAE-free, pixel-space Diffusion Transformer (DiT) architecture. Departing from text-to-image pretraining, the method integrates flow matching with a DINO similarity-based adaptive feature fusion mechanism to achieve efficient restoration. Despite possessing only 50 million parameters, the proposed model outperforms existing approaches in fidelity, perceptual quality, and robustness under single-step inference while significantly improving computational efficiency. These results establish a novel paradigm for lightweight, unified image inpainting, demonstrating that high-quality restoration can be achieved without reliance on heavy latent representations or large-scale generative pretraining.

Abstract

Unified image restoration (UIR) aims to recover high-quality (HQ) content from low-quality (LQ) images with different degradations using a single model. Most recent methods adapt large pretrained text-to-image (T2I) latent diffusion models for their strong capacity and generative priors. However, the variational autoencoder (VAE) in latent T2I models may discard restoration-sensitive details, while the open-ended synthesis prior can introduce content-inconsistent artifacts. We present PixRestore, a VAE-free pixel-space Diffusion Transformer (DiT) for UIR, where the diffusion backbone is trained entirely from scratch, without relying on T2I pretraining. PixRestore performs flow matching directly on patchified pixels, preserving fine-grained details while keeping the token sequence tractable. To adapt to different degradations, PixRestore learns to predict the reliability of layer features using LQ--HQ DINO feature similarity. Features from more reliable layers are fused as dense conditioning, while less reliable layers receive stronger HQ-feature supervision to encourage degradation removal. We train PixRestore on a large-scale corpus of diverse scenes and degradations, and further finetune it into a one-step generator using DINO-based adversarial objectives for efficient inference. Experiments on public benchmarks and real-world test sets show that, with only about 50M parameters and single-step inference, PixRestore achieves the best overall fidelity, perceptual quality, and robustness to degradations among competing UIR models while being far more efficient. Larger PixRestore variants can further boost performance, demonstrating the scalability of our pixel-space design. Code and the curated benchmark can be found at https://github.com/csslc/PixRestore.

Problem

• Unified Image Restoration
• Latent Diffusion Models
• VAE Detail Loss
• Content-Inconsistent Artifacts

Innovation

• Pixel-space Diffusion Transformer
• VAE-free
• Flow Matching
• DINO Feature Similarity
• One-step Generation