PixRestore: Unified Image Restoration via Pixel Diffusion Transformer

Summary

PixRestore is a VAE-free pixel-space Diffusion Transformer for unified image restoration (UIR), performing flow matching directly on patchified pixels to preserve fine details. It adapts to degradations by predicting layer feature reliability via LQ-HQ DINO similarity, fusing reliable layers as dense conditioning while supervising less reliable layers to remove degradation. Trained on a large-scale corpus and finetuned into a one-step generator with DINO-based adversarial objectives, PixRestore achieves the best overall fidelity, perceptual quality, and robustness among competing UIR models with only about 50M parameters and single-step inference, outperforming them in efficiency. Larger variants further boost performance, demonstrating scalability.