Efficient One-Step Diffusion Restoration Model with Compact Token Compression and Linear Attention

• URL: http://arxiv.org/abs/2605.23451v1
• Date: Fri, 22 May 2026 10:07:02 GMT
• Title: Efficient One-Step Diffusion Restoration Model with Compact Token Compression and Linear Attention
• Abstract summary: Existing Real-ISR methods inherit the dense latent representations and quadratic-cost global modeling paradigm.<n>We argue that the key bottleneck lies in excessive token redundancy and costly token interactions during high-resolution restoration.<n>We introduce a linear-attention DiT with LoRA fine-tuning, enabling efficient high-resolution restoration with linear-complexity token mixing.
• Score: 66.63806505114263
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Real-world image super-resolution aims to recover high-quality images from complex and unknown real-world degradations. However, existing generative Real-ISR methods largely inherit the dense latent representations and quadratic-cost global modeling paradigm developed for high-resolution image synthesis, causing computation, memory usage, and inference latency to scale unfavorably with resolution and thus limiting practical deployment. We argue that the key bottleneck lies not in insufficient restoration priors, but in excessive token redundancy and costly token interactions during high-resolution restoration. Motivated by this observation, we revisit Real-ISR from the perspectives of compact latent representation and linear-complexity modeling, and propose SANA-SR, an efficient one-step restoration framework. Specifically, SANA-SR employs a deep compression autoencoder with a 32x compression ratio to drastically reduce latent tokens while preserving restoration-relevant structures and textures. On top of this compact latent space, we introduce a linear-attention DiT with LoRA fine-tuning, enabling efficient high-resolution restoration with linear-complexity token mixing. Extensive experiments on all benchmark datasets demonstrate that SANA-SR achieves highly competitive and often superior quantitative performance against existing methods, while restoring clearer and more realistic textures. Moreover, after pruning, the deployed model runs in 0.019s with 407.95G MACs and 344M parameters, highlighting its strong potential for practical mobile deployment.

Related papers

• PixelIR: Fidelity-Perception Decoupling via Pixel-Space Image-Residual Flow Matching for Efficient One-Step Real-World Super-Resolution [28.787313933850687]
  Real-world image super-resolution (Real-ISR) aims to preserve structures supported by the degraded observation while reconstructing perceptually realistic details.<n>We propose PixelIR, a fidelity-perception decoupling framework built upon pixel-space image-residual flow matching.<n>Experiments show that PixelIR achieves leading PSNR, SSIM, and LPIPS on both RealSR and DRealSR.
  arXiv  Detail & Related papers  (2026-08-31T13:39:54Z)
• EchoSR: Efficient Context Harnessing for Lightweight Image Super-Resolution [22.100613250016366]
  EchoSR is an efficient context-harnessing framework for lightweight image super-resolution.<n>It unifies multi-scale receptive field modeling and hierarchical context fusion.<n>It consistently outperforms state-of-the-art lightweight super-resolution methods across multiple benchmarks.
  arXiv  Detail & Related papers  (2026-05-17T14:20:27Z)
• VARestorer: One-Step VAR Distillation for Real-World Image Super-Resolution [76.19751531910039]
  We propose VARestorer, a framework that transforms a pre-trained text-to-image VAR model into a one-step ISR model.<n>VARestorer achieves state-of-the-art performance with 72.32 MUSIQ and 0.7669 CLIPIQA on DIV2K dataset.
  arXiv  Detail & Related papers  (2026-04-23T09:09:36Z)
• ARCHE: Autoregressive Residual Compression with Hyperprior and Excitation [1.376408511310322]
  ARCHE is an end-to-end learned image compression framework that balances modeling accuracy and computational efficiency.<n>It attains state-of-the-art rate-distortion efficiency without relying on recurrent or transformer-based components.<n>Visual comparisons confirm sharper textures and improved color fidelity, particularly at lower bit rates.
  arXiv  Detail & Related papers  (2026-03-10T19:32:10Z)
• InfScene-SR: Spatially Continuous Inference for Arbitrary-Size Image Super-Resolution [3.6762434952581713]
  InfScene-SR is a framework enabling spatially continuous super-resolution for large, arbitrary scenes.<n>We adapt the iterative refinement process of diffusion models with a novel guided and variance-corrected fusion mechanism.
  arXiv  Detail & Related papers  (2026-02-23T11:34:59Z)
• Continuous Degradation Modeling via Latent Flow Matching for Real-World Super-Resolution [11.776067915986687]
  We introduce a novel framework capable of synthesizing authentic LR images from a single HR image by leveraging the latent degradation space with flow matching.<n>Our approach generates LR images with realistic artifacts at unseen degradation levels, which facilitates the creation of large-scale, real-world SR training datasets.
  arXiv  Detail & Related papers  (2026-02-04T04:16:38Z)
• SFTok: Bridging the Performance Gap in Discrete Tokenizers [72.9996757048065]
  We propose textbfSFTok, a discrete tokenizer that incorporates a multi-step iterative mechanism for precise reconstruction.<n>At a high compression rate of only 64 tokens per image, SFTok achieves state-of-the-art reconstruction quality on ImageNet.
  arXiv  Detail & Related papers  (2025-12-18T18:59:04Z)
• TinySR: Pruning Diffusion for Real-World Image Super-Resolution [35.07163534857897]
  We present TinySR, a compact yet effective diffusion model specifically designed for Real-ISR.<n>TinySR significantly reduces computational cost and model size, achieving up to 5.68x speedup and 83% parameter reduction compared to its teacher TSD-SR.
  arXiv  Detail & Related papers  (2025-08-24T16:17:33Z)
• RestoreVAR: Visual Autoregressive Generation for All-in-One Image Restoration [51.77917733024544]
  latent diffusion models (LDMs) have improved the perceptual quality of All-in-One image Restoration (AiOR) methods.<n>LDMs suffer from slow inference due to their iterative denoising process, rendering them impractical for time-sensitive applications.<n>Visual autoregressive modeling ( VAR) performs scale-space autoregression and achieves comparable performance to that of state-of-the-art diffusion transformers.
  arXiv  Detail & Related papers  (2025-05-23T15:52:26Z)
• ZipIR: Latent Pyramid Diffusion Transformer for High-Resolution Image Restoration [75.0053551643052]
  We introduce ZipIR, a novel framework that enhances efficiency, scalability, and long-range modeling for high-res image restoration.<n>ZipIR employs a highly compressed latent representation that compresses image 32x, effectively reducing the number of spatial tokens.<n>ZipIR surpasses existing diffusion-based methods, offering unmatched speed and quality in restoring high-resolution images from severely degraded inputs.
  arXiv  Detail & Related papers  (2025-04-11T14:49:52Z)
• RBSR: Efficient and Flexible Recurrent Network for Burst Super-Resolution [57.98314517861539]
  Burst super-resolution (BurstSR) aims at reconstructing a high-resolution (HR) image from a sequence of low-resolution (LR) and noisy images. In this paper, we suggest fusing cues frame-by-frame with an efficient and flexible recurrent network.
  arXiv  Detail & Related papers  (2023-06-30T12:14:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.