PixelSR improves image super-resolution by balancing detail and structure
PixelSR: Pixel-Space Orthogonal Flow Decoupling for Real-World Image Super-Resolution
Computer Vision and Pattern Recognition
Summary
Improving low-resolution images so they look realistic often causes the image’s important shapes and structures to get distorted. The authors show this trade-off is not a limitation of the main technology used, called diffusion models, but comes from the usual way these models are built. They introduce PixelSR, a new model that works directly on pixel data in a single step and separates semantic information from texture details. This approach helps make images both look realistic and keep correct structure.
What this means in practice
- •For photo editing software developers: Integrate PixelSR to enhance low-resolution photos with high-quality detail and accurate structure for better user experiences.$Commercial implications: Enables new or improved consumer photo editing products by providing superior image upscaling that preserves important image structures while adding realistic textures.
- •For video streaming service teams: Use PixelSR to upscale video frames in real time, maintaining visual detail and sharp edges for better streaming quality under bandwidth limits.
Authors
Xiang Lv, Kai Zhao, Qiang Wang, Qiao Mo, Chao Zhou, Mingwen Shao
Abstract
Latent diffusion models have recently achieved remarkable progress in Real-world Image Super-Resolution (Real-SR). However, existing methods consistently exhibit a persistent seesaw effect: enhancing perceptual realism compromises structural fidelity. This effect, we argue, is not intrinsic to diffusion models, but largely stems from the architectural constraints of prevailing two-stage frameworks. To address this challenge, we propose PixelSR, a novel single-stage Pixel Diffusion Transformer for Real-SR that operates directly in pixel space. Specifically, PixelSR consists of two core designs: Pixel-space Conditional Diffusion and Orthogonal Flow Decoupling. The former establishes an effective pixel-space restoration framework by integrating spatially aligned LR conditioning, visual semantic guidance, and lightweight texture refinement, thereby bypassing VAE compression and enabling end-to-end optimization directly in RGB space. Complementarily, the latter decomposes the asymmetric flow target into a low-rank semantic component and a noise-free orthogonal texture component. By routing these components to specialized branches, it achieves decoupled optimization with reduced mutual interference. Benefiting from these designs, our PixelSR restores perceptually realistic details while faithfully preserving structural content. Extensive experiments across multiple benchmarks demonstrate that PixelSR not only achieves competitive performance, but also effectively mitigates the fidelity--realism seesaw effect.