arxiv: 2604.11470 · v1 · submitted 2026-04-13 · 💻 cs.CV

Degradation-Aware and Structure-Preserving Diffusion for Real-World Image Super-Resolution

Yang Ji , Zonghao Chen , Zhihao Xue , Junqin Hu This is my paper

Pith reviewed 2026-05-10 16:48 UTC · model grok-4.3

classification 💻 cs.CV

keywords real-world image super-resolutiondiffusion modelsdegradation-aware restorationstructure preservationtoken injectionasymmetric noise injectionperceptual qualityimage restoration

0 comments

The pith

Degradation-aware token injection and edge-modulated noise let diffusion models restore real photos with better structure and realism.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Real-world super-resolution is difficult for diffusion models because actual photo degradations are complex and not explicitly modeled. The paper shows that two lightweight modules added to an existing diffusion setup can address this by encoding simple degradation statistics from the low-resolution input and mixing them into the conditioning features, plus adjusting the training noise according to local edge strength to protect important details. A reader would care because this approach could make AI enhancement of everyday images look more natural without requiring a complete redesign of the base model. If these modules deliver as described, diffusion methods would achieve competitive perceptual quality and a better balance between sharpness and natural appearance on real degraded photos. The reported experiments and ablations on standard benchmarks support that the modules work and complement each other.

Core claim

We propose a degradation-aware and structure-preserving diffusion framework for real-world SR. We introduce Degradation-aware Token Injection, which encodes lightweight degradation statistics from low-resolution inputs and fuses them with semantic conditioning features, enabling explicit degradation-aware restoration. We further propose Spatially Asymmetric Noise Injection, which modulates diffusion noise with local edge strength to better preserve structural regions during training. Both modules are lightweight add-ons to the adopted diffusion SR framework, requiring only minor modifications to the conditioning pipeline.

What carries the argument

Degradation-aware Token Injection and Spatially Asymmetric Noise Injection: two lightweight modules added to the diffusion conditioning pipeline that encode degradation statistics from the low-resolution input and modulate noise according to local edge strength.

If this is right

The method achieves competitive no-reference perceptual quality on real-world super-resolution benchmarks.
Restored images appear more realistic than those from recent diffusion baselines.
A favorable perception-distortion trade-off is maintained.
Ablation studies show that each module contributes and that the two together produce complementary improvements.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Explicit but lightweight degradation handling could extend to other generative image restoration tasks where full degradation modeling is impractical.
The edge-based noise modulation idea might apply to diffusion models in domains that require preserving fine details, such as medical or scientific imagery.
Because the modules require only small changes to the conditioning path, the approach could be combined with future improvements in base diffusion architectures without major retraining costs.

Load-bearing premise

That the gains in perceptual quality and structure preservation observed on the tested datasets arise from these two modules and will hold for other real-world degraded images.

What would settle it

A new test set of diverse real-world low-resolution images on which the method fails to match or exceed recent baselines in no-reference perceptual metrics and visual structure preservation would falsify the claim.

Figures

Figures reproduced from arXiv: 2604.11470 by Junqin Hu, Yang Ji, Zhihao Xue, Zonghao Chen.

**Figure 2.** Figure 2: Overview of the proposed framework. The DTI module extracts a six-dimensional degradation descriptor from the LR input [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Figure 3: Visual results of different methods on two typical real-world examples from the RealSR dataset. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

**Figure 4.** Figure 4: Representative failure case of the proposed method. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

read the original abstract

Real-world image super-resolution is particularly challenging for diffusion models because real degradations are complex, heterogeneous, and rarely modeled explicitly. We propose a degradation-aware and structure-preserving diffusion framework for real-world SR. Specifically, we introduce Degradation-aware Token Injection, which encodes lightweight degradation statistics from low-resolution inputs and fuses them with semantic conditioning features, enabling explicit degradation-aware restoration. We further propose Spatially Asymmetric Noise Injection, which modulates diffusion noise with local edge strength to better preserve structural regions during training. Both modules are lightweight add-ons to the adopted diffusion SR framework, requiring only minor modifications to the conditioning pipeline. Experiments on DIV2K and RealSR show that our method delivers competitive no-reference perceptual quality and visually more realistic restoration results than recent baselines, while maintaining a favorable perception--distortion trade-off. Ablations confirm the effectiveness of each module and their complementary gains when combined. The code and model are publicly available at https://github.com/jiyang0315/DASP-SR.git.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Two lightweight conditioning modules for diffusion SR deliver competitive perceptual results on standard benchmarks, but the gains look incremental and the tests do not strongly isolate generalization to heterogeneous real degradations.

read the letter

The main points are that the authors add Degradation-aware Token Injection to encode and fuse lightweight degradation stats from the LR input, plus Spatially Asymmetric Noise Injection that scales noise by local edge strength. Both are presented as simple add-ons to an existing diffusion SR backbone, and the abstract plus ablations claim they improve degradation handling and structure preservation while keeping the overhead low. The code is released, which helps, and the reported no-reference scores plus visual comparisons on DIV2K and RealSR look competitive with recent baselines along with a reasonable perception-distortion balance. Ablations show each module contributes and they combine positively, which is straightforward to follow.

Referee Report

2 major / 2 minor

Summary. The paper introduces a diffusion-based framework for real-world image super-resolution that adds two lightweight modules to an existing diffusion SR backbone: Degradation-aware Token Injection, which extracts and fuses lightweight degradation statistics from the LR input with semantic conditioning, and Spatially Asymmetric Noise Injection, which modulates the diffusion noise schedule according to local edge strength derived from the LR image. Experiments on DIV2K (synthetic) and RealSR (real) report competitive no-reference perceptual scores, subjectively more realistic outputs than recent baselines, and a favorable perception-distortion trade-off, with ablations indicating that each module contributes and that they combine favorably. Code and models are released publicly.

Significance. If the modules deliver robust, generalizable gains, the work would provide a practical way to make diffusion SR more explicitly degradation-aware and structure-preserving without substantial compute overhead. The public code release is a clear strength that supports reproducibility. The central claim, however, rests on whether the reported gains on DIV2K/RealSR reflect genuine improvements for heterogeneous real degradations rather than dataset-specific effects.

major comments (2)

[Experiments] Experiments section: the ablations demonstrate that each module improves results on DIV2K and RealSR and that they are complementary, yet no cross-dataset evaluation on additional real-world SR benchmarks (with degradation combinations outside the training distribution) is reported. This is load-bearing for the claim that the modules enable generalizable degradation-aware restoration, because the lightweight degradation statistics are extracted from the same training distribution used for the reported tables.
[Ablation study] Ablation study and results tables: no statistical significance tests, confidence intervals, or run-to-run variance are provided for the reported perceptual deltas (e.g., no-reference scores). Without these, it is difficult to determine whether the observed improvements exceed typical variance and therefore support the central claim of additive, robust gains from the two modules.

minor comments (2)

The abstract and method description would benefit from explicitly naming the no-reference metrics (e.g., NIQE, BRISQUE, or others) used to claim 'competitive no-reference perceptual quality.'
[Method] Figure captions and the description of Spatially Asymmetric Noise Injection could clarify how edge strength is computed from the LR input and whether any preprocessing is applied under heavy blur.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback and for recognizing the practical value of the proposed lightweight modules. We address each major comment below and describe the revisions we will make to strengthen the manuscript.

read point-by-point responses

Referee: [Experiments] Experiments section: the ablations demonstrate that each module improves results on DIV2K and RealSR and that they are complementary, yet no cross-dataset evaluation on additional real-world SR benchmarks (with degradation combinations outside the training distribution) is reported. This is load-bearing for the claim that the modules enable generalizable degradation-aware restoration, because the lightweight degradation statistics are extracted from the same training distribution used for the reported tables.

Authors: We agree that additional cross-dataset evaluations would further substantiate the generalizability of the Degradation-aware Token Injection module. While DIV2K and RealSR are standard benchmarks covering synthetic and real degradations, we will add quantitative and qualitative results on at least one additional real-world SR benchmark (e.g., the DRealSR dataset) whose degradation characteristics differ from the training distribution. These new experiments will be reported in the revised Experiments section together with an updated discussion of generalization. revision: yes
Referee: [Ablation study] Ablation study and results tables: no statistical significance tests, confidence intervals, or run-to-run variance are provided for the reported perceptual deltas (e.g., no-reference scores). Without these, it is difficult to determine whether the observed improvements exceed typical variance and therefore support the central claim of additive, robust gains from the two modules.

Authors: We acknowledge that the current ablation tables report single-run perceptual scores without variance estimates or statistical tests. In the revised manuscript we will rerun the key ablation configurations across multiple random seeds, report mean and standard deviation for the no-reference metrics, and include confidence intervals. Where appropriate we will also add a brief note on statistical significance to confirm that the observed additive gains are robust. revision: yes

Circularity Check

0 steps flagged

No circularity in proposed modules or empirical claims

full rationale

The paper proposes two lightweight add-on modules (Degradation-aware Token Injection and Spatially Asymmetric Noise Injection) to an existing diffusion SR framework and validates them through standard ablations plus quantitative/qualitative comparisons on DIV2K and RealSR. No mathematical derivation chain, equations, or first-principles results are presented that reduce to self-definitions, fitted inputs renamed as predictions, or self-citation load-bearing premises. All claims rest on external dataset benchmarks and module ablations rather than internal consistency loops or renamed known patterns. This is a typical empirical ML contribution with no detectable circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that the two proposed modules meaningfully address complex real degradations and that standard benchmarks suffice to demonstrate superiority; no free parameters or invented entities are explicitly introduced in the abstract.

axioms (2)

domain assumption Lightweight degradation statistics extracted from low-resolution inputs can be effectively fused into semantic conditioning features for improved restoration.
Core premise of the Degradation-aware Token Injection module.
domain assumption Modulating diffusion noise according to local edge strength during training preserves structural details without harming overall generation quality.
Core premise of the Spatially Asymmetric Noise Injection module.

pith-pipeline@v0.9.0 · 5476 in / 1176 out tokens · 40644 ms · 2026-05-10T16:48:14.083350+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages

[1]

Pda-rwsr: Pixel-wise degradation adaptive real-world super-resolution

Andreas Aakerberg, Majed El Helou, Kamal Nasrollahi, and Thomas Moeslund. Pda-rwsr: Pixel-wise degradation adaptive real-world super-resolution. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 4097–4107, 2024. 2

work page 2024
[2]

Pda-rwsr: Pixel-wise degradation adaptive real-world super-resolution

Andreas Aakerberg, Majed El Helou, Kamal Nasrollahi, and Thomas Moeslund. Pda-rwsr: Pixel-wise degradation adaptive real-world super-resolution. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 4097–4107, 2024. 3

work page 2024
[3]

Ntire 2017 challenge on single image super-resolution: Dataset and study

Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. InPro- ceedings of the IEEE conference on computer vision and pat- tern recognition workshops, pages 126–135, 2017. 2, 6

work page 2017
[4]

The perception-distortion tradeoff

Yochai Blau and Tomer Michaeli. The perception-distortion tradeoff. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 6228–6237, 2018. 2, 6

work page 2018
[5]

The 2018 pirm challenge on percep- tual image super-resolution

Yochai Blau, Roey Mechrez, Radu Timofte, Tomer Michaeli, and Lihi Zelnik-Manor. The 2018 pirm challenge on percep- tual image super-resolution. InProceedings of the European conference on computer vision (ECCV) workshops, pages 0– 0, 2018. 2, 6

work page 2018
[6]

Toward real-world single image super-resolution: A new benchmark and a new model

Jianrui Cai, Hui Zeng, Hongwei Yong, Zisheng Cao, and Lei Zhang. Toward real-world single image super-resolution: A new benchmark and a new model. InProceedings of the IEEE/CVF international conference on computer vision, pages 3086–3095, 2019. 2, 6

work page 2019
[7]

The Fourth Challenge on Image Super-Resolution (×4) at NTIRE 2026: Bench- mark Results and Method Overview

Zheng Chen, Kai Liu, Jingkai Wang, Xianglong Yan, Jianze Li, Ziqing Zhang, Jue Gong, Jiatong Li, Lei Sun, Xiaoyang Liu, Radu Timofte, Yulun Zhang, et al. The Fourth Challenge on Image Super-Resolution (×4) at NTIRE 2026: Bench- mark Results and Method Overview. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Work...

work page 2026
[8]

Image super-resolution using deep convolutional net- works.IEEE transactions on pattern analysis and machine intelligence, 38(2):295–307, 2015

Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image super-resolution using deep convolutional net- works.IEEE transactions on pattern analysis and machine intelligence, 38(2):295–307, 2015. 2

work page 2015
[9]

Image restoration by denoising diffusion models with iteratively preconditioned guidance

Tomer Garber and Tom Tirer. Image restoration by denoising diffusion models with iteratively preconditioned guidance. InProceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 25245–25254, 2024. 3

work page 2024
[10]

Denoising dif- fusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020. 2, 3

work page 2020
[11]

Denoising diffusion restoration models.Advances in neural information processing systems, 35:23593–23606,

Bahjat Kawar, Michael Elad, Stefano Ermon, and Jiaming Song. Denoising diffusion restoration models.Advances in neural information processing systems, 35:23593–23606,

work page
[12]

Musiq: Multi-scale image quality transformer

Junjie Ke, Qifei Wang, Yilin Wang, Peyman Milanfar, and Feng Yang. Musiq: Multi-scale image quality transformer. InProceedings of the IEEE/CVF international conference on computer vision, pages 5148–5157, 2021. 2, 6

work page 2021
[13]

Accurate image super-resolution using very deep convolutional net- works

Jiwon Kim, Jung Kwon Lee, and Kyoung Mu Lee. Accurate image super-resolution using very deep convolutional net- works. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 1646–1654, 2016. 2

work page 2016
[14]

Photo- realistic single image super-resolution using a generative ad- versarial network

Christian Ledig, Lucas Theis, Ferenc Husz´ar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, et al. Photo- realistic single image super-resolution using a generative ad- versarial network. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4681–4690,

work page
[15]

Blinddiff: empowering degradation modeling in diffusion models for blind image super-resolution.Science China Information Sciences, 69 (1):112102, 2026

Feng Li, Yixuan Wu, Zichao Liang, Runmin Cong, Huihui Bai, Yao Zhao, and Meng Wang. Blinddiff: empowering degradation modeling in diffusion models for blind image super-resolution.Science China Information Sciences, 69 (1):112102, 2026. 3

work page 2026
[16]

Learning heterogeneous degradation representation for real- world super-resolution

Haowei Li, Pengxu Wei, Dongyu Zhang, and Liang Lin. Learning heterogeneous degradation representation for real- world super-resolution. InThe Fourteenth International Conference on Learning Representations. 3

work page
[17]

Srdiff: Single image super-resolution with diffusion probabilistic models

Haoying Li, Yifan Yang, Meng Chang, Shiqi Chen, Huajun Feng, Zhihai Xu, Qi Li, and Yueting Chen. Srdiff: Single image super-resolution with diffusion probabilistic models. Neurocomputing, 479:47–59, 2022. 2

work page 2022
[18]

Lsdir: A large scale dataset for image restoration

Yawei Li, Kai Zhang, Jingyun Liang, Jiezhang Cao, Ce Liu, Rui Gong, Yulun Zhang, Hao Tang, Yun Liu, Denis Deman- dolx, et al. Lsdir: A large scale dataset for image restoration. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1775–1787, 2023. 6

work page 2023
[19]

Enhanced deep residual networks for single image super-resolution

Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual networks for single image super-resolution. InProceedings of the IEEE confer- ence on computer vision and pattern recognition workshops, pages 136–144, 2017. 2

work page 2017
[20]

Diff- bir: Toward blind image restoration with generative diffusion prior

Xinqi Lin, Jingwen He, Ziyan Chen, Zhaoyang Lyu, Bo Dai, Fanghua Yu, Yu Qiao, Wanli Ouyang, and Chao Dong. Diff- bir: Toward blind image restoration with generative diffusion prior. InEuropean conference on computer vision, pages 430–448. Springer, 2024. 2, 3, 6, 7

work page 2024
[21]

Cd- former: When degradation prediction embraces diffusion model for blind image super-resolution

Qingguo Liu, Chenyi Zhuang, Pan Gao, and Jie Qin. Cd- former: When degradation prediction embraces diffusion model for blind image super-resolution. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7455–7464, 2024. 3

work page 2024
[22]

Image restoration with mean- reverting stochastic differential equations

Ziwei Luo, Fredrik K Gustafsson, Zheng Zhao, Jens Sj¨olund, and Thomas B Sch ¨on. Image restoration with mean- reverting stochastic differential equations.arXiv preprint arXiv:2301.11699, 2023. 3

work page arXiv 2023
[23]

completely blind

Anish Mittal, Rajiv Soundararajan, and Alan C Bovik. Mak- ing a “completely blind” image quality analyzer.IEEE Sig- nal processing letters, 20(3):209–212, 2012. 2, 6

work page 2012
[24]

Improved denoising diffusion probabilistic models

Alexander Quinn Nichol and Prafulla Dhariwal. Improved denoising diffusion probabilistic models. InInternational conference on machine learning, pages 8162–8171. PMLR,

work page
[25]

High-resolution image synthesis with latent diffusion models

Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 2

work page 2022
[26]

Image super- resolution via iterative refinement.IEEE transactions on pattern analysis and machine intelligence, 45(4):4713–4726,

Chitwan Saharia, Jonathan Ho, William Chan, Tim Sali- mans, David J Fleet, and Mohammad Norouzi. Image super- resolution via iterative refinement.IEEE transactions on pattern analysis and machine intelligence, 45(4):4713–4726,

work page
[27]

Pixel-level and semantic-level adjustable super-resolution: A dual-lora approach

Lingchen Sun, Rongyuan Wu, Zhiyuan Ma, Shuaizheng Liu, Qiaosi Yi, and Lei Zhang. Pixel-level and semantic-level adjustable super-resolution: A dual-lora approach. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2333–2343, 2025. 6, 7

work page 2025
[28]

Ex- ploring clip for assessing the look and feel of images

Jianyi Wang, Kelvin CK Chan, and Chen Change Loy. Ex- ploring clip for assessing the look and feel of images. InPro- ceedings of the AAAI conference on artificial intelligence, pages 2555–2563, 2023. 2, 6

work page 2023
[29]

Exploiting diffusion prior for real-world image super-resolution.International Journal of Computer Vision, 132(12):5929–5949, 2024

Jianyi Wang, Zongsheng Yue, Shangchen Zhou, Kelvin CK Chan, and Chen Change Loy. Exploiting diffusion prior for real-world image super-resolution.International Journal of Computer Vision, 132(12):5929–5949, 2024. 2, 3, 6, 7

work page 2024
[30]

Esrgan: En- hanced super-resolution generative adversarial networks

Xintao Wang, Ke Yu, Shixiang Wu, Jinjin Gu, Yihao Liu, Chao Dong, Yu Qiao, and Chen Change Loy. Esrgan: En- hanced super-resolution generative adversarial networks. In Proceedings of the European conference on computer vision (ECCV) workshops, pages 0–0, 2018. 3

work page 2018
[31]

Real-esrgan: Training real-world blind super-resolution with pure synthetic data

Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super-resolution with pure synthetic data. InProceedings of the IEEE/CVF inter- national conference on computer vision, pages 1905–1914,

work page 1905
[32]

Sinsr: diffusion-based image super- resolution in a single step

Yufei Wang, Wenhan Yang, Xinyuan Chen, Yaohui Wang, Lanqing Guo, Lap-Pui Chau, Ziwei Liu, Yu Qiao, Alex C Kot, and Bihan Wen. Sinsr: diffusion-based image super- resolution in a single step. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 25796–25805, 2024. 3

work page 2024
[33]

Image quality assessment: from error visibility to structural similarity.IEEE transactions on image processing, 13(4):600–612, 2004

Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Si- moncelli. Image quality assessment: from error visibility to structural similarity.IEEE transactions on image processing, 13(4):600–612, 2004. 2, 6

work page 2004
[34]

One-step effective diffusion network for real-world image super-resolution.Advances in Neural Information Process- ing Systems, 37:92529–92553, 2024

Rongyuan Wu, Lingchen Sun, Zhiyuan Ma, and Lei Zhang. One-step effective diffusion network for real-world image super-resolution.Advances in Neural Information Process- ing Systems, 37:92529–92553, 2024. 6, 7

work page 2024
[35]

Seesr: Towards semantics- aware real-world image super-resolution

Rongyuan Wu, Tao Yang, Lingchen Sun, Zhengqiang Zhang, Shuai Li, and Lei Zhang. Seesr: Towards semantics- aware real-world image super-resolution. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 25456–25467, 2024. 2, 3, 6, 7

work page 2024
[36]

Pixel-aware stable diffusion for realistic image super-resolution and personalized stylization

Tao Yang, Rongyuan Wu, Peiran Ren, Xuansong Xie, and Lei Zhang. Pixel-aware stable diffusion for realistic image super-resolution and personalized stylization. InEuropean conference on computer vision, pages 74–91. Springer, 2024. 3, 6, 7

work page 2024
[37]

Learning diffusion texture priors for image restoration

Tian Ye, Sixiang Chen, Wenhao Chai, Zhaohu Xing, Jing Qin, Ge Lin, and Lei Zhu. Learning diffusion texture priors for image restoration. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 2524–2534, 2024. 3

work page 2024
[38]

Resshift: Efficient diffusion model for image super- resolution by residual shifting.Advances in neural infor- mation processing systems, 36:13294–13307, 2023

Zongsheng Yue, Jianyi Wang, and Chen Change Loy. Resshift: Efficient diffusion model for image super- resolution by residual shifting.Advances in neural infor- mation processing systems, 36:13294–13307, 2023. 3, 6, 7

work page 2023
[39]

Arbitrary-steps image super-resolution via diffusion inver- sion

Zongsheng Yue, Kang Liao, and Chen Change Loy. Arbitrary-steps image super-resolution via diffusion inver- sion. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 23153–23163, 2025. 6, 7

work page 2025
[40]

Designing a practical degradation model for deep blind im- age super-resolution

Kai Zhang, Jingyun Liang, Luc Van Gool, and Radu Timofte. Designing a practical degradation model for deep blind im- age super-resolution. InProceedings of the IEEE/CVF inter- national conference on computer vision, pages 4791–4800,

work page
[41]

The unreasonable effectiveness of deep features as a perceptual metric

Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shecht- man, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 6

work page 2018