Pith. sign in

REVIEW 4 major objections 4 minor 2 cited by

RFSR: Improving ISR Diffusion Models via Reward Feedback Learning

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Fine-tuning diffusion-based super-resolution models with timestep-aware reward feedback learning improves their perceptual and aesthetic output quality, as shown by higher no-reference quality scores on several established models.

desk verdict A sensible plug-and-play reward fine-tuning recipe for diffusion-based ISR with a coherent timestep-aware design, but the central subjective-quality claim rests on metrics that the method itself optimizes and that the paper's own ablation shows can diverge from visual quality. read the letter →

arxiv 2412.03268 v1 pith:DCB2AYWE submitted 2024-12-04 cs.CV

Xiaopeng Sun , Qinwei Lin , Yu Gao , Yujie Zhong , Chengjian Feng , Dengjie Li , Zheng Zhao , Jie Hu
show 1 more author
Lin Ma
This is my paper · ORCID
classification cs.CV
keywords imagesuper-resolutiondiffusionmodelsrewardfeedbacklearningtimestep-awaretrainingCLIP-IQAGram-KLregularizationperceptualquality
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that diffusion-based image super-resolution models, typically trained with denoising losses, can be further improved by fine-tuning them with reward feedback learning. The proposed method, RFSR, splits the denoising trajectory: early steps are constrained on low-frequency structure to keep the image faithful, while later steps are optimized with aesthetic and perceptual rewards to boost subjective quality. A Gram-KL regularizer is added to prevent the stylization artifacts that pure reward optimization causes. Experiments on DiffBIR, PASD, and SeeSR show consistent gains on no-reference perceptual and aesthetic metrics, suggesting the approach works as a plug-and-play improvement layer.

What carries the argument

The method uses three components. A low-frequency structure constraint, computed with the discrete wavelet transform, matches the LL subband of early denoising outputs to the ground truth, keeping the overall layout stable. Reward feedback learning at late timesteps uses CLIP-IQA and ImageReward as differentiable reward models, pushing outputs toward higher perceptual and human-preference scores. A Gram-KL regularizer penalizes differences between VGG Gram matrices of the fine-tuned and frozen pretrained models, opposing the stylistic shifts characteristic of reward hacking. Gradients are updated only at the final denoising step to avoid instability, and the whole procedure fine-tunes an already-trained model in a plug-and-play fashion.

What would settle it

A blind user study comparing original and RFSR-fine-tuned outputs of DiffBIR, PASD, and SeeSR on diverse real-world images, scored by human raters, would settle the claim; if human preference does not track the reported MANIQA, CLIPIQA, and Aesthetic gains, the core assumption fails. An additional holdout set of images from a distribution unlike the training data would test whether the reward models' judgments generalize.

Watch

Extended reading notes

Core claim

The central claim is that reward feedback learning, selectively applied to the later denoising steps of a diffusion-based super-resolution model, can push output images toward higher perceptual and aesthetic quality while preserving the structural fidelity established early in the denoising process. The authors support this by showing, via discrete wavelet transform analysis, that low-frequency structure is settled early while high-frequency texture develops later and tends to diverge from the ground truth. They therefore apply a low-frequency constraint at large timesteps and a reward loss at small timesteps, which outperforms applying either constraint uniformly. For example, fine-tuning SeeSR with RFSR raises MANIQA from 0.5091 to 0.5954 and CLIPIQA from 0.6989 to 0.7944 on DIV2K-val.

Load-bearing premise

The reward models CLIP-IQA and ImageReward are reliable, differentiable proxies for human perceptual quality when used as training losses, so that optimizing them actually improves how people perceive the super-resolved images.

Editorial extensions

If this is right

  • Existing diffusion-based ISR models can be upgraded with RFSR without architectural changes or retraining from scratch.
  • The timestep-aware split — structure constraint early, reward late — offers a general recipe for fine-tuning other conditional diffusion restoration models.
  • The Gram-KL regularizer provides a lightweight counter to reward hacking that is orthogonal to LoRA or KL-based constraints.
  • The reported metric gains indicate that perceptual quality can be improved while keeping fidelity (LPIPS) close to the original model's level.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because CLIPIQA is used both as a training reward and as the headline evaluation metric, part of the measured gain is by construction; a blind human-preference study would be needed to confirm the improvements are genuinely perceptual.
  • The timestep-aware recipe could transfer to other conditional generation tasks such as inpainting or deblurring, where early structure and late texture also separate.
  • The divergence of high-frequency details from ground truth in late denoising steps suggests a fundamental tension between fidelity metrics and perceived quality; reward fine-tuning deliberately trades off the former for the latter.
  • The Gram-KL regularizer depends on VGG feature statistics; testing whether other feature extractors or style statistics (mean/covariance) behave similarly would clarify the mechanism.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes RFSR, a plug-and-play fine-tuning method for diffusion-based image super-resolution (ISR) models. During early denoising steps the method imposes a low-frequency DWT constraint against the ground truth to preserve structure; during later steps it trains with reward feedback from CLIP-IQA and ImageReward, plus a Gram-KL regularization intended to mitigate reward hacking. Experiments on DIV2K-val, DRealSR, and RealSR report improved MANIQA, MUSIQ, CLIPIQA, and Aesthetic scores for DiffBIR, PASD, and SeeSR after RFSR fine-tuning. The paper claims 'excellent subjective results' based on these metrics and selected visual comparisons, and releases code.

Significance. If the subjective-quality claim were convincingly established, RFSR would be a useful general recipe: it is model-agnostic, simple to implement, and the code is released. The authors also deserve credit for identifying and attempting to mitigate reward hacking in the ISR setting via Gram-KL regularization. However, the central evidence is weakened by a circularity: CLIP-IQA is both a training reward (Eq. 5) and a headline evaluation metric (Table 1), and Table 4 explicitly shows a reward-hacked baseline with higher CLIPIQA and Aesthetic scores than the proposed method. The claim of 'excellent subjective results' therefore currently rests on selected images and non-reference metrics rather than on demonstrated human preference. The significance is conditional on adding a human evaluation and more rigorous statistical reporting.

major comments (4)
  1. [Sec. 3.3, Eq. (5); Sec. 4.2, Metrics] CLIP-IQA is used as a training reward in Eq. (5) and then reported as the primary evaluation metric in Tables 1, 2, 3, and 4. Part of the reported CLIPIQA gain is therefore by construction, and the same applies, to a smaller extent, to ImageReward-aligned aesthetic judgments. To support the claim that RFSR improves perceptual quality, the paper should either report a human preference study, or evaluate on metrics that are not directly or indirectly optimized by the training rewards, or both.
  2. [Sec. 4.4, Table 4; Sec. 1, Fig. 2] Table 4 shows that the 'w/o regularization' baseline achieves CLIPIQA 0.8964 and Aesthetic 5.3612, which are substantially higher than the proposed Gram-KL method's 0.7944 and 5.2683, yet the authors describe that baseline as reward hacking with visibly degraded images. This is direct evidence that the reported reward metrics can be inversely related to subjective quality. The paper's own data therefore undermine the inference from Table 1's metric gains to 'excellent subjective results'; a human evaluation is needed to establish the central claim.
  3. [Sec. 4, Tables 1-5] All quantitative results are reported from a single training run with no error bars, no multiple seeds, and no statistical significance tests. Some differences between configurations are small (e.g., Aesthetic 5.2683 vs. 5.2669 in Table 4), so it is not possible to tell whether the reported improvements are robust. The authors should provide at least 3 seeds, report mean and standard deviation, and discuss checkpoint selection or early stopping, especially since Fig. 2 shows metric behavior varying strongly with training iterations.
  4. [Sec. 4.3, Qualitative Comparisons] The qualitative claim that RFSR 'excels at enhancing high-quality texture details' is supported only by selected crops in Figure 4. Since the paper explicitly acknowledges that reward models 'lack robustness when confronted with larger-scale real-world data and diffusion-generated data' (Sec. 5), the subjective evidence should be supplemented with a formal user study, ideally with multiple raters and a forced-choice protocol against the baseline models.
minor comments (4)
  1. [Sec. 3.2, Eq. (2)] The notation in Eq. (2) uses absolute-value bars, which should be clarified as an L1 norm; otherwise the equation is dimensionally ambiguous.
  2. [Sec. 3.2, Eq. (1)] The definition of DWT(·)_LL should be made explicit; the text refers to 'DWT(It)_LL' but Eq. (1) defines only the full DWT output.
  3. [Sec. 4.4, Table 2] The row labels in Table 2 are inconsistent with the text: the paper's described default setting 'st1=20, st2=40' is listed only as 'Ours', while the first row 'st1∈[1,40], st2∈[41,50]' merges two different interval lengths. Please clarify which rows correspond to which sampling-step schedules.
  4. [Sec. 1, Contributions] The claim 'We are the first to introduce reward feedback learning into super-resolution fine-tuning' is strong and should be positioned more carefully against existing reward-finetuning works for diffusion models [4, 6, 35] and any prior use of perceptual rewards in restoration.

Circularity Check

1 steps flagged · score 6.0 of 10

The CLIPIQA gain reported as perceptual-quality evidence is the direct training objective of Eq. (5), so part of the central claim is circular.

  1. fitted input called prediction [Section 3.3 Eq. (5); Section 4.2 metrics; Section 4.3 and Table 1]
    "In the later denoising stages, we use reward feedback learning to improve the perceptual and aesthetic quality of the SR images. ... Lreward = L(RW (ct, It)) = λclipiqaLCLIP-IQA(It) + λiwLIW(ct, It). ... the methods fine-tuned with RFSR achieve significant improvements in both perceptual and subjective metrics. For example, on the DRealSR dataset, PASD-RFSR achieves ... 37% over CLIPIQA."

    CLIPIQA is the function being maximized by the reward loss in Eq. (5). Reporting the post-fine-tuning CLIPIQA increase as evidence of 'perceptual and subjective' improvement is not an independent measurement; the increase is the expected result of gradient ascent on that same reward. The paper itself shows CLIPIQA can rise while subjective quality degrades ('reward hacking' in Table 4: CLIPIQA 0.8964, Aesthetic 5.3612 for the no-regularization baseline), so the optimized metric cannot by itself establish the subjective-quality claim. MANIQA, MUSIQ, and Aesthetic are distinct no-reference functions and give partial independent evidence, but the central 'excellent subjective results' claim has no human study and thus rests in part on the very reward used for training.

full rationale

The method is otherwise self-contained: the low-frequency DWT constraint, timestep-aware schedule, and Gram-KL regularization are defined independently of the evaluation metrics, and there are no load-bearing self-citations. The circular element is the use of CLIP-IQA both as the training reward (Eq. 5) and as a headline evaluation metric (Tables 1, 3, 4). Optimizing CLIPIQA and then reporting higher CLIPIQA is partly a statement of successful optimization rather than independent confirmation of perceptual quality. The paper's own Figure 2 and Table 4 acknowledge that CLIPIQA can increase while images become visibly stylized, so the metric is not a trustworthy proxy for the 'excellent subjective results' claim. A human preference study would be needed to confirm the subjective part; without it, the central perceptual-quality claim is partially circular and partially under-supported, but not entirely forced, because several other metrics and qualitative comparisons are reported.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The method's central claim depends on the reliability of two external reward models, the wavelet-domain decomposition of structure, and an unverified gradient-truncation scheme. Loss weights and timestep boundaries are hand-chosen hyperparameters. No new physical or conceptual entities are introduced.

free parameters (5)
  • lambda_dwt = 0.0005
    Weight for the low-frequency DWT loss; chosen by authors without a reported search.
  • lambda_clipiqa = 0.00005
    Weight for the CLIP-IQA reward loss.
  • lambda_iw = 0.000005
    Weight for the ImageReward loss.
  • lambda_r = 0.000005
    Weight for the Gram-KL regularizer.
  • st1_st2_intervals = st1=20, st2=40 for DiffBIR/SeeSR; st1=8, st2=17 for PASD
    Timestep boundaries for early vs. late loss schedules, set per model based on their inference steps.
assumptions (5)
  • domain assumption CLIP-IQA and ImageReward scores correlate with human perceptual and aesthetic judgments of super-resolved images.
    The training objective assumes these models are reliable rewards; the paper's limitation section acknowledges they lack robustness on diffusion-generated data.
  • domain assumption The low-frequency component of an image, extracted via DWT, captures the structural information that should be preserved in early denoising steps.
    Equation 2 and Figure 1 motivate this split; it is a modeling choice based on observations of one model (SeeSR) and is assumed to transfer.
  • domain assumption Gram matrix distances in VGG feature space measure style differences and can suppress reward-hacking stylization.
    Equation 6 follows style transfer literature; the paper calls it 'Gram-KL' though it is an L2 distance, not a KL divergence.
  • ad hoc to paper Backpropagating rewards only through the final denoising step (instead of the whole trajectory) is sufficient to fine-tune the model and avoids gradient explosion.
    Section 3.5 states this without derivation; the gradient path for early-step low-frequency loss is not fully specified.
  • domain assumption The Real-ESRGAN degradation pipeline used to synthesize training pairs is representative of real-world degradations.
    Section 4.2; the method is evaluated on real-world sets after training only on synthetic pairs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RFSR: Improving ISR Diffusion Models via Reward Feedback Learning." pith.science (2026). https://pith.science/paper/DCB2AYWE

@misc{pith2026241203268,
  author       = {Pith},
  title        = {Pith review of: RFSR: Improving ISR Diffusion Models via Reward Feedback Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DCB2AYWE}},
  note         = {Machine review of arXiv:2412.03268}
}
read the original abstract

Generative diffusion models (DM) have been extensively utilized in image super-resolution (ISR). Most of the existing methods adopt the denoising loss from DDPMs for model optimization. We posit that introducing reward feedback learning to finetune the existing models can further improve the quality of the generated images. In this paper, we propose a timestep-aware training strategy with reward feedback learning. Specifically, in the initial denoising stages of ISR diffusion, we apply low-frequency constraints to super-resolution (SR) images to maintain structural stability. In the later denoising stages, we use reward feedback learning to improve the perceptual and aesthetic quality of the SR images. In addition, we incorporate Gram-KL regularization to alleviate stylization caused by reward hacking. Our method can be integrated into any diffusion-based ISR model in a plug-and-play manner. Experiments show that ISR diffusion models, when fine-tuned with our method, significantly improve the perceptual and aesthetic quality of SR images, achieving excellent subjective results. Code: https://github.com/sxpro/RFSR

Figures

Figures reproduced from arXiv: 2412.03268 by the authors.

Figure 1
Figure 1. (a) The first row shows the progressive denoising of the image during the iterative process, while the next two rows show the [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Visualization for Reward Hacking. The direct application of reward feedback learning significantly improves the perceptual [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overview of our method. 3.3. Reward Feedback Learning To significantly improve the subjective performance of the super-resolution model, we introduce reward feed￾back learning to fine-tune the parameters θ in the super￾resolution model G. Unlike most diffusion methods, which refine predictions sequentially from the last step xT to the initial step x ′ 0 (xT → xT −1 → · · · → x ′ 0 ), we adopt an innovative approach … view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: A visual comparison of state-of-the-art ISR diffusion models and their counterparts trained with our RFSR is presented. Each [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Effectiveness of Timestep-Aware Training. An exces [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Effectiveness of Reward Feedback Models. The intro [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 8
Figure 8. Figure 8: Effectiveness of Low-Frequency Constraints. ISR gen [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 7
Figure 7. Figure 7: Effectiveness of Style Regularization.Without regular [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Zoom-IQA: Image Quality Assessment with Reliable Region-Aware Reasoning

    cs.CV 2026-01 conditional novelty 7.0 of 10

    Zoom-IQA lets a vision-language model iteratively crop and zoom into image regions before giving a quality score, improving reasoning and restoration guidance over single-pass IQA models.

  2. TASR: Timestep-Aware Diffusion Model for Image Super-Resolution

    cs.CV 2024-12 conditional novelty 5.0 of 10

    A timestep-aware adapter that weights ControlNet guidance early and pretrained diffusion features late improves no-reference perceptual quality in image super-resolution.

Reference graph

Works this paper leans on

41 extracted references · 17 canonical work pages · cited by 2 Pith papers

  1. [1]

    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. In Pro- ceedings of the IEEE conference on computer vision and pat- tern recognition workshops, pages 126–135, 2017. 6

    2017

  2. [2]

    Training diffusion models with reinforce- ment learning

    Kevin Black, Michael Janner, Yilun Du, Ilya Kostrikov, and Sergey Levine. Training diffusion models with reinforce- ment learning. arXiv preprint arXiv:2305.13301, 2023. 3

    arXiv 2023

  3. [3]

    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. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3086–3095, 2019. 6

    work page 2019

  4. [4]

    Directly fine-tuning diffusion models on differentiable re- wards

    Kevin Clark, Paul Vicol, Kevin Swersky, and David J Fleet. Directly fine-tuning diffusion models on differentiable re- wards. arXiv preprint arXiv:2309.17400, 2023. 2, 3, 4, 5, 7

    arXiv 2023

  5. [5]

    Raft: Reward ranked finetuning for generative foundation model alignment

    Hanze Dong, Wei Xiong, Deepanshu Goyal, Yihan Zhang, Winnie Chow, Rui Pan, Shizhe Diao, Jipeng Zhang, Kashun Shum, and Tong Zhang. Raft: Reward ranked finetuning for generative foundation model alignment. arXiv preprint arXiv:2304.06767, 2023. 1

    arXiv 2023

  6. [6]

    Re- inforcement learning for fine-tuning text-to-image diffusion models

    Ying Fan, Olivia Watkins, Yuqing Du, Hao Liu, Moonkyung Ryu, Craig Boutilier, Pieter Abbeel, Moham- mad Ghavamzadeh, Kangwook Lee, and Kimin Lee. Re- inforcement learning for fine-tuning text-to-image diffusion models. Advances in Neural Information Processing Sys- tems, 36, 2024. 4, 8

    work page 2024

  7. [7]

    Im- age style transfer using convolutional neural networks

    Leon A Gatys, Alexander S Ecker, and Matthias Bethge. Im- age style transfer using convolutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2414–2423, 2016. 4

    work page 2016

  8. [8]

    Div8k: Diverse 8k resolution image dataset

    Shuhang Gu, Andreas Lugmayr, Martin Danelljan, Manuel Fritsche, Julien Lamour, and Radu Timofte. Div8k: Diverse 8k resolution image dataset. In 2019 IEEE/CVF Interna- tional Conference on Computer Vision Workshop (ICCVW), pages 3512–3516. IEEE, 2019. 6

    work page 2019

Show all 41 references
  1. [9]

    Denoising dif- fusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020. 1

    2020

  2. [10]

    A style-based generator architecture for generative adversarial networks

    Tero Karras, Samuli Laine, and Timo Aila. A style-based generator architecture for generative adversarial networks. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 4401–4410, 2019. 6

    2019

  3. [11]

    Diffu- sionclip: Text-guided diffusion models for robust image ma- nipulation

    Gwanghyun Kim, Taesung Kwon, and Jong Chul Ye. Diffu- sionclip: Text-guided diffusion models for robust image ma- nipulation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2426–2435,

  4. [12]

    Adam: A method for stochastic opti- mization

    Diederik P Kingma. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980, 2014. 5

    2014 arXiv

  5. [13]

    Pick-a-pic: An open dataset of user preferences for text-to-image generation

    Yuval Kirstain, Adam Polyak, Uriel Singer, Shahbuland Ma- tiana, Joe Penna, and Omer Levy. Pick-a-pic: An open dataset of user preferences for text-to-image generation. Advances in Neural Information Processing Systems , 36: 36652–36663, 2023. 3

    2023

  6. [14]

    Aligning text- to-image models using human feedback

    Kimin Lee, Hao Liu, Moonkyung Ryu, Olivia Watkins, Yuqing Du, Craig Boutilier, Pieter Abbeel, Mohammad Ghavamzadeh, and Shixiang Shane Gu. Aligning text- to-image models using human feedback. arXiv preprint arXiv:2302.12192, 2023. 1

    2023 arXiv

  7. [15]

    Controlnet++: Improving conditional controls with efficient consistency feedback

    Ming Li, Taojiannan Yang, Huafeng Kuang, Jie Wu, Zhaon- ing Wang, Xuefeng Xiao, and Chen Chen. Controlnet++: Improving conditional controls with efficient consistency feedback. In European Conference on Computer Vision , pages 129–147. Springer, 2025. 1

    2025

  8. [16]

    Diff- bir: Towards blind image restoration with generative diffu- sion prior

    Xinqi Lin, Jingwen He, Ziyan Chen, Zhaoyang Lyu, Bo Dai, Fanghua Yu, Wanli Ouyang, Yu Qiao, and Chao Dong. Diff- bir: Towards blind image restoration with generative diffu- sion prior. arXiv preprint arXiv:2308.15070, 2023. 1, 2, 5

    2023 arXiv

  9. [17]

    Aligning text-to-image diffusion models with reward backpropagation

    Mihir Prabhudesai, Anirudh Goyal, Deepak Pathak, and Katerina Fragkiadaki. Aligning text-to-image diffusion models with reward backpropagation. arXiv preprint arXiv:2310.03739, 2023. 3

    2023 arXiv

  10. [18]

    Xpsr: Cross-modal priors for diffusion-based image super-resolution

    Yunpeng Qu, Kun Yuan, Kai Zhao, Qizhi Xie, Jinhua Hao, Ming Sun, and Chao Zhou. Xpsr: Cross-modal priors for diffusion-based image super-resolution. In European Con- ference on Computer Vision, pages 285–303. Springer, 2025. 2

    2025

  11. [19]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  12. [20]

    Byteedit: Boost, comply and acceler- ate generative image editing

    Yuxi Ren, Jie Wu, Yanzuo Lu, Huafeng Kuang, Xin Xia, Xionghui Wang, Qianqian Wang, Yixing Zhu, Pan Xie, Shiyin Wang, et al. Byteedit: Boost, comply and acceler- ate generative image editing. In European Conference on Computer Vision, pages 184–200. Springer, 2025. 1

    2025

  13. [21]

    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. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022. 1

    2022

  14. [22]

    Photorealistic text-to-image diffusion models with deep language understanding

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily L Denton, Kamyar Ghasemipour, Raphael Gontijo Lopes, Burcu Karagol Ayan, Tim Salimans, et al. Photorealistic text-to-image diffusion models with deep language understanding. Advances in neural information...

    2022

  15. [23]

    Laion-5b: An open large-scale dataset for training next generation image-text models

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Worts- man, et al. Laion-5b: An open large-scale dataset for training next generation image-text models. Advances in Neural In- f...

    2022

  16. [24]

    Freeu: Free lunch in diffusion u-net

    Chenyang Si, Ziqi Huang, Yuming Jiang, and Ziwei Liu. Freeu: Free lunch in diffusion u-net. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4733–4743, 2024. 1

    2024

  17. [25]

    Very deep convo- lutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convo- lutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014. 5 9

    2014 arXiv

  18. [26]

    https://stability.ai/stablediffusion

    Stability.ai. https://stability.ai/stablediffusion. 2

  19. [27]

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

    Peiran Ren Xuansong Xie Tao Yang, Rongyuan Wu and Lei Zhang. Pixel-aware stable diffusion for realistic image super- resolution and personalized stylization. In The European Conference on Computer Vision (ECCV) 2024, 2023. 1, 2, 5

    2024

  20. [28]

    Ntire 2017 challenge on single image super-resolution: Methods and results

    Radu Timofte, Eirikur Agustsson, Luc Van Gool, Ming- Hsuan Yang, and Lei Zhang. Ntire 2017 challenge on single image super-resolution: Methods and results. In Proceed- ings of the IEEE conference on computer vision and pattern recognition workshops, pages 114–125, 2017. 6

    2017

  21. [29]

    Chan, and Chen Change Loy

    Jianyi Wang, Zongsheng Yue, Shangchen Zhou, Kelvin C.K. Chan, and Chen Change Loy. Exploiting diffusion prior for real-world image super-resolution. 2024. 1, 2

    2024

  22. [30]

    Recovering realistic texture in image super-resolution by deep spatial feature transform

    Xintao Wang, Ke Yu, Chao Dong, and Chen Change Loy. Recovering realistic texture in image super-resolution by deep spatial feature transform. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 606–615, 2018. 6

    2018

  23. [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. In Proceedings of the IEEE/CVF inter- national conference on computer vision , pages 1905–1914,

    1905

  24. [32]

    Image quality assessment: from error visibility to structural similarity

    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. 1

    2004

  25. [33]

    Component divide- and-conquer for real-world image super-resolution

    Pengxu Wei, Ziwei Xie, Hannan Lu, Zongyuan Zhan, Qixi- ang Ye, Wangmeng Zuo, and Liang Lin. Component divide- and-conquer for real-world image super-resolution. In Com- puter Vision–ECCV 2020: 16th European Conference, Glas- gow, UK, August 23–28, 2020, Proceedings, Part VIII ...

    2020

  26. [34]

    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. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 25456–25467, 2024. 1, 2, 5

    2024

  27. [35]

    Imagere- ward: Learning and evaluating human preferences for text- to-image generation

    Jiazheng Xu, Xiao Liu, Yuchen Wu, Yuxuan Tong, Qinkai Li, Ming Ding, Jie Tang, and Yuxiao Dong. Imagere- ward: Learning and evaluating human preferences for text- to-image generation. Advances in Neural Information Pro- cessing Systems, 36, 2024. 1, 3, 4, 5

    2024

  28. [36]

    Maniqa: Multi-dimension attention network for no-reference image quality assessment

    Sidi Yang, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, and Yujiu Yang. Maniqa: Multi-dimension attention network for no-reference image quality assessment. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pag...

    2022

  29. [37]

    A dense reward view on aligning text-to-image diffusion with prefer- ence

    Shentao Yang, Tianqi Chen, and Mingyuan Zhou. A dense reward view on aligning text-to-image diffusion with prefer- ence. arXiv preprint arXiv:2402.08265, 2024. 1

    2024 arXiv

  30. [38]

    Self-play fine-tuning of diffusion models for text-to-image generation

    Huizhuo Yuan, Zixiang Chen, Kaixuan Ji, and Quanquan Gu. Self-play fine-tuning of diffusion models for text-to-image generation. arXiv preprint arXiv:2402.10210, 2024. 1

    2024 arXiv

  31. [39]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 3836–3847, 2023. 4

    2023

  32. [40]

    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. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 586–595, 2018. 6

    2018

  33. [41]

    Confronting reward overoptimiza- tion for diffusion models: A perspective of inductive and pri- macy biases

    Ziyi Zhang, Sen Zhang, Yibing Zhan, Yong Luo, Yonggang Wen, and Dacheng Tao. Confronting reward overoptimiza- tion for diffusion models: A perspective of inductive and pri- macy biases. arXiv preprint arXiv:2402.08552, 2024. 1 10

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.