Pith. sign in

REVIEW 4 major objections 5 minor 92 references

After the Party: Navigating the Mapping From Color to Ambient Lighting

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper argues that ambient lighting normalization can be extended from single white light to multiple colored lights, and offers the CL3AN dataset and the RLN2 network as the means to do it.

desk verdict A promising dataset idea whose credibility hinges on a capture protocol the draft doesn't provide—worth refereeing, but the reviewer should demand the calibration details. read the letter →

arxiv 2508.02168 v2 pith:VHOHB7WW submitted 2025-08-04 cs.CV

Florin-Alexandru Vasluianu , Tim Seizinger , Zongwei Wu , Radu Timofte This is my paper
classification cs.CV
keywords AmbientLightingNormalizationcoloredlightsourcesmulti-illuminantscenesillumination-reflectancedecompositionRetinexpaireddatasetchromaticity-luminanceguidanceimagerestoration
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 tries to establish that Ambient Lighting Normalization can handle multiple colored light sources, not just the single white or white-balanced lighting assumed by most prior work. It introduces CL3AN, a dataset of paired captures in which the same scenes are recorded under multiple colored lights and under uniform ambient light, and claims this is the first large-scale, high-resolution dataset built for that task. Alongside the dataset, it proposes RLN2, a learning framework that restores the ambient-lit image by explicitly guiding a Retinex-style separation of illumination from reflectance using chromaticity and luminance components. If these claims hold, CL3AN provides a benchmark that exposes the failures of existing methods—illumination inconsistencies, texture leakage, and color distortion—and RLN2 offers a way to remove colored lighting artifacts at competitive computational cost.

What carries the argument

The load-bearing objects are the CL3AN capture protocol and the RLN2 network. In CL3AN, each scene is photographed once under multiple colored (RGB) light sources and once under uniform ambient light; the direct-lighting setup removes the color consistency constraint of earlier datasets so that complex material-light interactions appear. RLN2 then learns the mapping between the two captures by explicit chromaticity (color) and luminance (brightness) component guidance, a Retinex-inspired instruction that forces the network to separate illumination from reflectance rather than memorize a global color transform. That explicit decomposition is what the paper says lets the model avoid the artifacts seen in existing methods.

What would settle it

Run dense correspondence between the colored-light and ambient versions of the same CL3AN scene: if static parts of the scene show residual geometric motion, or if flat patches of known color differ beyond illumination, the paired ground truth is not valid. That failure would falsify the dataset and the comparisons built on it.

Watch

Extended reading notes

Core claim

The paper's central claim is that colored, multi-source lighting can be normalized to an ambient-lit reference without sacrificing robustness or speed, provided the model is told how to separate what the lights do from what the surfaces look like. The evidence offered is CL3AN, a large-scale, high-resolution paired dataset in which direct lighting is produced by RGB lights and the ambient reference is acquired under a separate uniform lighting setup, deliberately dropping the color consistency constraint used by earlier datasets. On top of it, the paper presents RLN2, which uses explicit chromaticity-luminance component guidance, inspired by the Retinex model, to perform the illumination-reflectance decomposition needed for restoration. According to the paper, benchmarking shows that leading approaches produce artifacts because they cannot disentangle illumination from reflectance, while RLN2 handles non-homogeneous color lighting and material-specific reflectance variations with competitive computational cost.

Load-bearing premise

The central claim depends on the CL3AN paired captures: the colored-light and ambient-lit images of each scene must be pixel-aligned, and the ambient image must be a correct, lighting-independent ground truth for the same scene, yet the provided text does not show the capture, alignment, and post-processing details that would verify this.

Editorial extensions

If this is right

  • Ambient Lighting Normalization can be evaluated under multiple colored light sources, not just single white or white-aligned lighting, making the task closer to real indoor and event scenes.
  • Existing restoration models trained on single-light or white-domain data can be measured and shown to produce illumination inconsistencies, texture leakage, and color distortion on CL3AN.
  • RLN2 can serve as a preprocessing step for applications that need illumination-invariant inputs, such as neural image editing, so that downstream editing or recognition sees reflectances rather than colored shadows.
  • Because RLN2 stays computationally competitive, colored-light normalization can be applied in practical settings where diffusion-based restoration is too slow.

Reading between the lines

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

  • If CL3AN's pairing is sound, it could become a shared testbed for neighboring problems such as color constancy and white balance under multiple illuminants, since it provides controlled color-shifted inputs with known ambient references.
  • A natural extension would be to record CL3AN-style pairs as video or under varying light directions, turning the static normalization task into a relighting benchmark; the paper does not attempt this.
  • The explicit chromaticity-luminance guidance suggests that smaller, non-generative models may close much of the gap with diffusion-based restoration on color-dominated degradations, which would be a testable hypothesis on other restoration tasks.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The manuscript proposes CL3AN, described as the first large-scale, high-resolution dataset for Ambient Lighting Normalization (ALN) under multiple colored light sources, together with RLN2, a learning framework that performs illumination-reflectance decomposition guided by explicit chromaticity-luminance components. The paper argues that existing datasets and methods oversimplify illumination by assuming single or white-balanced light sources, and that CL3AN addresses this gap by using RGB direct lighting without color-consistency constraints. The provided text contains the abstract, introduction, the beginning of Section 3, and references, but no experimental section, dataset statistics, evaluation tables, or complete method details. The central claims are therefore not verifiable from the manuscript as presented.

Significance. If the claims hold, CL3AN would be a valuable new benchmark for a realistic and under-addressed problem, and RLN2 would offer a competitive solution at moderate computational cost. The paper identifies a genuine limitation of prior datasets such as AMBIENT6K and proposes a plausible direction via colored-light direct lighting. The stated intention to release code, models, and benchmark data is a positive contribution. However, the current manuscript provides no evidence for the headline claims: no dataset statistics, no capture validation, no evaluation protocol, no results, and no complete method description. The significance cannot be assessed until these are supplied.

major comments (4)
  1. [Section 1, Figure 2(D)] The load-bearing assumption of CL3AN is that the colored-light input and the ambient-lit reference are pixel-aligned captures of the same scene under identical camera settings. The text states that the direct lighting setup 'is based on RGB lights, dropping the color consistency constraint,' but it does not specify the camera, exposure, aperture, white balance, tone mapping, or any alignment procedure used to obtain the paired images. Nor does it address effects such as specular highlights, interreflections, or sensor saturation that the ambient reference cannot represent. Without this capture protocol and alignment validation, the ground truth for the benchmark is not well-defined and every downstream comparison loses meaning.
  2. [Abstract and full text] The abstract claims 'Extensive evaluations on existing benchmarks and our dataset demonstrate the effectiveness of our approach' and 'highly competitive computational cost,' but the supplied text contains no experimental section, no tables, no metrics, no dataset statistics, no ablations, and no evaluation protocol. The paper cannot be assessed for soundness until these are provided. Please include dataset size and resolution, number of scenes and lighting configurations, train/test splits, evaluation metrics, comparison methods, and runtime or FLOPs measurements.
  3. [Section 3] Section 3 begins with the sentence 'The core of our work is extending the study of Ambient Lighting Normalization to direct color lighting' and then jumps to a figure caption and an incomplete paragraph starting 'Provided statistics, such as.' The actual RLN2 architecture, the 'explicit chromaticity-luminance components guidance,' the loss functions, and the training details are absent. It is therefore impossible to evaluate the novelty of the method, to verify that the claimed decomposition is actually learned, or to reproduce the approach. Please provide a complete method section with equations, a network diagram, and training hyperparameters.
  4. [Abstract and Figure 2] The claim that CL3AN is 'the first large-scale, high-resolution dataset of its kind' is unsupported by any concrete numbers. The text does not define what 'large-scale' and 'high-resolution' mean in this context, nor does it compare the dataset size, resolution, or diversity with existing benchmarks such as ISTD/ISTD+, WSRD, AMBIENT6K, and LSMI. Additionally, since WSRD and AMBIENT6K are from the same group as the current paper, the relationship and independence of the new benchmark should be clarified when reporting comparisons on those datasets.
minor comments (5)
  1. [Overall structure] The manuscript is missing Section 2 (related work is apparently absent) and the numbering jumps from Section 1 to Section 3; this should be fixed in a complete version.
  2. [Page 4 paragraph] The paragraph beginning 'Provided statistics, such as' is an incomplete sentence and the statistics it refers to are not defined; please complete the thought or remove the fragment.
  3. [Author affiliation] The affiliation contains 'W¨urzburg' with an umlaut encoding error; it should be typeset as 'Würzburg.'
  4. [Figure 2] The four subfigures (A)-(D) are described in the caption, but the text does not consistently refer to all of them; please add explicit cross-references in the text.
  5. [References] Reference [67] is cited as 'A Vaswani' with an incomplete author list; please use the full 'Vaswani et al.' citation for 'Attention is all you need.'

Circularity Check

0 steps flagged · score 2.0 of 10

Minor self-citations in task positioning and baselines (AMBIENT6K, WSRD), but no circular prediction/fit loop; RLN2 is a supervised mapping to an independently collected paired target.

full rationale

The available text contains no derivation chain in which a predicted quantity is defined in terms of the same fitted quantity. The central empirical contribution is CL3AN, a paired capture dataset whose input is a scene under RGB colored light and whose target is an ambient-lit reference (Fig. 2(D) caption: 'the direct lighting setup is based on RGB lights, dropping the color consistency constraint'). RLN2 is trained to regress that target; supervised learning against a collected ground truth is not a prediction-from-fit loop. The Retinex reference (Land 1964) is an external source of architectural inspiration, not an imported uniqueness theorem. The self-cited AMBIENT6K [66] and WSRD [65] are used to position the task and as benchmarks/baselines; the statement that AMBIENT6K is limited ('the strong connection between the operating parameters set for the direct and ambient lighting systems limits the study to exposure correction for white-aligned scene lighting') is a description of the authors' own prior dataset rather than a load-bearing citation that forces the present result. No exhibited reduction of any claim to its inputs is present. One support gap is real but not circular: the provided excerpt omits the CL3AN capture, alignment, and exposure-matching protocol, so the validity of the paired ground truth cannot be checked from this text; that is a completeness risk, not a circularity.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

The available text does not reveal hand-tuned constants or explicit free parameters; the network weights are learned from data rather than fixed by the authors. The central claims rest on two domain assumptions: that Retinex-style decomposition into chromaticity and luminance is sufficient for this task, and that the CL3AN paired capture produces valid, aligned ground-truth ambient references. No new physical entities, such as new forces or particles, are introduced.

assumptions (2)
  • domain assumption Images under colored lighting can be decomposed into reflectance (chromaticity) and illumination (luminance) components that can be recombined to yield an ambient-normalized image.
    This Retinex assumption is stated in the abstract and Section 3 as the design principle behind RLN2.
  • domain assumption The CL3AN capture setup yields pixel-aligned pairs in which the ambient image is a correct, lighting-independent ground truth for the colored-light input.
    The dataset is the central claim; the validity of its paired ground truth is asserted in the dataset description (Fig. 2(D)) but the capture and alignment details are not in the provided text.

how reviews work

0 comments
Cite this review

Pith. "Pith review of After the Party: Navigating the Mapping From Color to Ambient Lighting." pith.science (2026). https://pith.science/paper/VHOHB7WW

@misc{pith2026250802168,
  author       = {Pith},
  title        = {Pith review of: After the Party: Navigating the Mapping From Color to Ambient Lighting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VHOHB7WW}},
  note         = {Machine review of arXiv:2508.02168}
}
read the original abstract

Illumination in practical scenarios is inherently complex, involving colored light sources, occlusions, and diverse material interactions that produce intricate reflectance and shading effects. However, existing methods often oversimplify this challenge by assuming a single light source or uniform, white-balanced lighting, leaving many of these complexities unaddressed. In this paper, we introduce CL3AN, the first large-scale, high-resolution dataset of its kind designed to facilitate the restoration of images captured under multiple Colored Light sources to their Ambient-Normalized counterparts. Through benchmarking, we find that leading approaches often produce artifacts, such as illumination inconsistencies, texture leakage, and color distortion, primarily due to their limited ability to precisely disentangle illumination from reflectance. Motivated by this insight, we achieve such a desired decomposition through a novel learning framework that leverages explicit chromaticity-luminance components guidance, drawing inspiration from the principles of the Retinex model. Extensive evaluations on existing benchmarks and our dataset demonstrate the effectiveness of our approach, showcasing enhanced robustness under non-homogeneous color lighting and material-specific reflectance variations, all while maintaining a highly competitive computational cost. The benchmark, codes, and models are available at www.github.com/fvasluianu97/RLN2.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

92 extracted references · 66 canonical work pages

  1. [1]

    Deep white-balance editing

    Mahmoud Afifi and Michael S Brown. Deep white-balance editing. In Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, pages 1397–1406,

  2. [2]

    Cross-camera convolutional color constancy

    Mahmoud Afifi, Jonathan T Barron, Chloe LeGendre, Yun- Ta Tsai, and Francois Bleibel. Cross-camera convolutional color constancy. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision , pages 1981–1990,

    1981

  3. [3]

    Arniqa: Learning distortion mani- fold for image quality assessment

    Lorenzo Agnolucci, Leonardo Galteri, Marco Bertini, and Alberto Del Bimbo. Arniqa: Learning distortion mani- fold for image quality assessment. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 189–198, 2024. 2

    2024

  4. [4]

    Towards Real-World Focus Stacking with Deep Learning

    Alexandre Araujo, Jean Ponce, and Julien Mairal. Towards real-world focus stacking with deep learning. arXiv preprint arXiv:2311.17846, 2023. 2

    work page Pith review arXiv 2023

  5. [5]

    2d ob- ject recognition: a comparative analysis of sift, surf and orb feature descriptors

    Monika Bansal, Munish Kumar, and Manish Kumar. 2d ob- ject recognition: a comparative analysis of sift, surf and orb feature descriptors. Multimedia Tools and Applications, 80 (12):18839–18857, 2021. 2

    2021

  6. [6]

    Convolutional color constancy

    Jonathan T Barron. Convolutional color constancy. In Pro- ceedings of the IEEE International Conference on Computer Vision, pages 379–387, 2015. 2

    2015

  7. [7]

    Shape, illumination, and reflectance from shading

    Jonathan T Barron and Jitendra Malik. Shape, illumination, and reflectance from shading. IEEE transactions on pattern analysis and machine intelligence , 37(8):1670–1687, 2014. 1

    2014

  8. [8]

    User-guided white balance for mixed lighting con- ditions

    Ivaylo Boyadzhiev, Kavita Bala, Sylvain Paris, and Fr ´edo Durand. User-guided white balance for mixed lighting con- ditions. ACM Trans. Graph., 31(6):200–1, 2012. 2

    2012

Show all 92 references
  1. [9]

    Retinexformer: One-stage retinex- based transformer for low-light image enhancement

    Yuanhao Cai, Hao Bian, Jing Lin, Haoqian Wang, Radu Tim- ofte, and Yulun Zhang. Retinexformer: One-stage retinex- based transformer for low-light image enhancement. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12504–12513, 2023. 2, 3, 6, 7

    2023

  2. [10]

    Simple baselines for image restoration

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration. arXiv preprint arXiv:2204.04676, 2022. 3, 6, 7

    2022 arXiv

  3. [11]

    Inout: Diverse image outpainting via gan inversion

    Yen-Chi Cheng, Chieh Hubert Lin, Hsin-Ying Lee, Jian Ren, Sergey Tulyakov, and Ming-Hsuan Yang. Inout: Diverse image outpainting via gan inversion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11431–11440, 2022. 2

    2022

  4. [12]

    Selective frequency network for image restoration

    Yuning Cui, Yi Tao, Zhenshan Bing, Wenqi Ren, Xinwei Gao, Xiaochun Cao, Kai Huang, and Alois Knoll. Selective frequency network for image restoration. InThe Eleventh In- ternational Conference on Learning Representations , 2023. 2, 3, 6, 7

    2023

  5. [13]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. 3

    2010 arXiv

  6. [14]

    Alpha- pose: Whole-body regional multi-person pose estimation and tracking in real-time

    Hao-Shu Fang, Jiefeng Li, Hongyang Tang, Chao Xu, Haoyi Zhu, Yuliang Xiu, Yong-Lu Li, and Cewu Lu. Alpha- pose: Whole-body regional multi-person pose estimation and tracking in real-time. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(6):7157–7173, 2022. 2

    2022

  7. [15]

    Finlayson, Steven D

    Graham D. Finlayson, Steven D. Hordley, and Paul M. Hubel. Color by correlation: A simple, unifying framework for color constancy. IEEE Transactions on Pattern Analysis and Machine Intelligence, 23(11):1209–1221, 2001. 2

    2001

  8. [16]

    Color constancy

    David H Foster. Color constancy. Vision research, 51(7): 674–700, 2011. 2

    2011

  9. [17]

    Building rome on a cloudless day

    Jan-Michael Frahm, Pierre Fite-Georgel, David Gallup, Tim Johnson, Rahul Raguram, Changchang Wu, Yi-Hung Jen, Enrique Dunn, Brian Clipp, Svetlana Lazebnik, and Marc Pollefeys. Building rome on a cloudless day. In Computer Vision – ECCV 2010 , pages 368–381, Berlin, Heidelberg,

    2010

  10. [18]

    Dw-gan: A discrete wavelet transform gan for nonho- mogeneous dehazing

    Minghan Fu, Huan Liu, Yankun Yu, Jun Chen, and Keyan Wang. Dw-gan: A discrete wavelet transform gan for nonho- mogeneous dehazing. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (CVPR) Workshops, pages 203–212, 2021. 3

    2021

  11. [19]

    Guiding instruction-based im- age editing via multimodal large language models

    Tsu-Jui Fu, Wenze Hu, Xianzhi Du, William Yang Wang, Yinfei Yang, and Zhe Gan. Guiding instruction-based im- age editing via multimodal large language models. arXiv preprint arXiv:2309.17102, 2023. 1, 8

    2023 arXiv

  12. [20]

    Color constancy for multiple light sources

    Arjan Gijsenij, Rui Lu, and Theo Gevers. Color constancy for multiple light sources. IEEE Transactions on image pro- cessing, 21(2):697–707, 2011. 2

    2011

  13. [21]

    Mamba: Linear-time sequence modeling with selective state spaces

    Albert Gu and Tri Dao. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752, 2023. 3, 6

    2023 arXiv

  14. [22]

    Pct-net: Full resolution image harmonization using pixel-wise color transformations

    Julian Jorge Andrade Guerreiro, Mitsuru Nakazawa, and Bj¨orn Stenger. Pct-net: Full resolution image harmonization using pixel-wise color transformations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5917–5926, 2023. 2

    2023

  15. [23]

    Mambair: A simple baseline for im- age restoration with state-space model

    Hang Guo, Jinmin Li, Tao Dai, Zhihao Ouyang, Xudong Ren, and Shu-Tao Xia. Mambair: A simple baseline for im- age restoration with state-space model. In European Confer- ence on Computer Vision, pages 222–241. Springer, 2025. 2, 3, 7

    2025

  16. [24]

    Asic: Aligning sparse in-the-wild image collections

    Kamal Gupta, Varun Jampani, Carlos Esteves, Abhinav Shri- vastava, Ameesh Makadia, Noah Snavely, and Abhishek Kar. Asic: Aligning sparse in-the-wild image collections. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4134–4145, 2023. 2

    2023

  17. [25]

    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. 3

    2020

  18. [26]

    Mask-shadowgan: Learning to remove shadows from unpaired data

    Xiaowei Hu, Yitong Jiang, Chi-Wing Fu, and Pheng-Ann Heng. Mask-shadowgan: Learning to remove shadows from unpaired data. In Proceedings of the IEEE/CVF interna- tional conference on computer vision , pages 2472–2481,

  19. [27]

    Range scaling global u-net for perceptual image enhance- ment on mobile devices

    Jie Huang, Pengfei Zhu, Mingrui Geng, Jiewen Ran, Xing- guang Zhou, Chen Xing, Pengfei Wan, and Xiangyang Ji. Range scaling global u-net for perceptual image enhance- ment on mobile devices. In Proceedings of the European conference on computer vision (ECCV) workshops, pages 0...

    2018

  20. [28]

    Image shadow removal via multi-scale deep retinex de- composition

    Yan Huang, Xinchang Lu, Yuhui Quan, Yong Xu, and Hui Ji. Image shadow removal via multi-scale deep retinex de- composition. Pattern Recognition, page 111126, 2024. 3

    2024

  21. [29]

    Illuminant spectra-based source separa- tion using flash photography

    Zhuo Hui, Kalyan Sunkavalli, Sunil Hadap, and Aswin C Sankaranarayanan. Illuminant spectra-based source separa- tion using flash photography. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 6209–6218, 2018. 3

    2018

  22. [30]

    Learning to separate multiple illuminants in a single image

    Zhuo Hui, Ayan Chakrabarti, Kalyan Sunkavalli, and Aswin C Sankaranarayanan. Learning to separate multiple illuminants in a single image. In Computer Vision and Pat- tern Recognition (CVPR 2019), 2019. 3

    2019

  23. [31]

    Texture feature extraction methods: A survey

    Anne Humeau-Heurtier. Texture feature extraction methods: A survey. IEEE access, 7:8975–9000, 2019. 1

    2019

  24. [32]

    Replac- ing mobile camera isp with a single deep learning model

    Andrey Ignatov, Luc Van Gool, and Radu Timofte. Replac- ing mobile camera isp with a single deep learning model. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops , pages 536–537,

  25. [33]

    Oneformer: One transformer to rule universal image segmentation

    Jitesh Jain, Jiachen Li, Mang Tik Chiu, Ali Hassani, Nikita Orlov, and Humphrey Shi. Oneformer: One transformer to rule universal image segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2989–2998, 2023. 1

    2023

  26. [34]

    Low-light image enhancement using gamma correction prior in mixed color spaces

    Jong Ju Jeon, Jun Young Park, and Il Kyu Eom. Low-light image enhancement using gamma correction prior in mixed color spaces. Pattern Recognition, 146:110001, 2024. 2

    2024

  27. [35]

    Neural gaffer: Relighting any object via diffusion.Advances in Neu- ral Information Processing Systems , 37:141129–141152,

    Haian Jin, Yuan Li, Fujun Luan, Yuanbo Xiangli, Sai Bi, Kai Zhang, Zexiang Xu, Jin Sun, and Noah Snavely. Neural gaffer: Relighting any object via diffusion.Advances in Neu- ral Information Processing Systems , 37:141129–141152,

  28. [36]

    Des3: Adaptive attention-driven self and soft shadow removal using vit similarity

    Yeying Jin, Wei Ye, Wenhan Yang, Yuan Yuan, and Robby T Tan. Des3: Adaptive attention-driven self and soft shadow removal using vit similarity. arXiv preprint arXiv:2211.08089, 2022. 1

    2022 arXiv

  29. [37]

    Hinet: Deep image hiding by invertible network

    Junpeng Jing, Xin Deng, Mai Xu, Jianyi Wang, and Zhenyu Guan. Hinet: Deep image hiding by invertible network. In Proceedings of the IEEE/CVF international conference on computer vision, pages 4733–4742, 2021. 3, 6, 7

    2021

  30. [38]

    Large scale multi-illuminant (lsmi) dataset for developing white balance algorithm under mixed illumination

    Dongyoung Kim, Jinwoo Kim, Seonghyeon Nam, Dongwoo Lee, Yeonkyung Lee, Nahyup Kang, Hyong-Euk Lee, Byun- gIn Yoo, Jae-Joon Han, and Seon Joo Kim. Large scale multi-illuminant (lsmi) dataset for developing white balance algorithm under mixed illumination. In Proceedings of the ...

    2021

  31. [39]

    Efficient frequency domain-based trans- formers for high-quality image deblurring

    Lingshun Kong, Jiangxin Dong, Jianjun Ge, Mingqiang Li, and Jinshan Pan. Efficient frequency domain-based trans- formers for high-quality image deblurring. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 5886–5895, 2023. 3

    2023

  32. [40]

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural net- works. In Proceedings of the 25th International Conference on Neural Information Processing Systems - Volume 1, page 1097–1105, Red Hook, NY , USA, 2012. Curran...

    2012

  33. [41]

    Lipsync3d: Data-efficient learning of per- sonalized 3d talking faces from video using pose and light- ing normalization

    Avisek Lahiri, Vivek Kwatra, Christian Frueh, John Lewis, and Chris Bregler. Lipsync3d: Data-efficient learning of per- sonalized 3d talking faces from video using pose and light- ing normalization. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern rec...

    2021

  34. [42]

    Video stitching for linear cam- era arrays

    Wei-Sheng Lai, Orazio Gallo, Jinwei Gu, Deqing Sun, Ming- Hsuan Yang, and Jan Kautz. Video stitching for linear cam- era arrays. arXiv preprint arXiv:1907.13622, 2019. 2

    1907 arXiv

  35. [43]

    EDWIN H. LAND. The retinex. American Scientist, 52(2): 247–264, 1964. 3, 5

    1964

  36. [44]

    Physics-based shadow im- age decomposition for shadow removal

    Hieu Le and Dimitris Samaras. Physics-based shadow im- age decomposition for shadow removal. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(12):9088– 9101, 2021. 1, 3

    2021

  37. [45]

    Mimt: Multi-illuminant color constancy via multi- task local surface and light color learning

    Shuwei Li, Jikai Wang, Michael S Brown, and Robby T Tan. Mimt: Multi-illuminant color constancy via multi- task local surface and light color learning. arXiv preprint arXiv:2211.08772, 2022. 2

    2022 arXiv

  38. [46]

    Discrete cosin transformer: Image modeling from frequency domain

    Xinyu Li, Yanyi Zhang, Jianbo Yuan, Hanlin Lu, and Yibo Zhu. Discrete cosin transformer: Image modeling from frequency domain. In Proceedings of the IEEE/CVF Win- ter Conference on Applications of Computer Vision , pages 5468–5478, 2023. 3

    2023

  39. [47]

    Effi- cient and explicit modelling of image hierarchies for image restoration

    Yawei Li, Yuchen Fan, Xiaoyu Xiang, Denis Demandolx, Rakesh Ranjan, Radu Timofte, and Luc Van Gool. Effi- cient and explicit modelling of image hierarchies for image restoration. In Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition, 2023. 2, 3, 6, 7

    2023

  40. [48]

    Swinir: Image restoration us- ing swin transformer

    Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, Luc Van Gool, and Radu Timofte. Swinir: Image restoration us- ing swin transformer. InProceedings of the IEEE/CVF inter- national conference on computer vision , pages 1833–1844,

  41. [49]

    Vrt: A video restoration transformer

    Jingyun Liang, Jiezhang Cao, Yuchen Fan, Kai Zhang, Rakesh Ranjan, Yawei Li, Radu Timofte, and Luc Van Gool. Vrt: A video restoration transformer. IEEE Transactions on Image Processing, 33:2171–2182, 2024. 3

    2024

  42. [50]

    A convnet for the 2020s

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feicht- enhofer, Trevor Darrell, and Saining Xie. A convnet for the 2020s. In Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition , pages 11976–11986,

  43. [51]

    Refusion: Enabling large-size realis- tic image restoration with latent-space diffusion models

    Ziwei Luo, Fredrik K Gustafsson, Zheng Zhao, Jens Sj¨olund, and Thomas B Sch ¨on. Refusion: Enabling large-size realis- tic image restoration with latent-space diffusion models. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 1680–...

    2023

  44. [52]

    U-mamba: Enhancing long-range dependency for biomedical image segmentation

    Jun Ma, Feifei Li, and Bo Wang. U-mamba: Enhancing long-range dependency for biomedical image segmentation. arXiv preprint arXiv:2401.04722, 2024. 1

    2024 arXiv

  45. [53]

    Color correction for tone mapping

    Radoslaw Mantiuk, Rafal Mantiuk, Anna Tomaszewska, and Wolfgang Heidrich. Color correction for tone mapping. In Computer graphics forum, pages 193–202. Wiley Online Li- brary, 2009. 2

    2009

  46. [54]

    Image seg- mentation using deep learning: A survey

    Shervin Minaee, Yuri Boykov, Fatih Porikli, Antonio Plaza, Nasser Kehtarnavaz, and Demetri Terzopoulos. Image seg- mentation using deep learning: A survey. IEEE transactions on pattern analysis and machine intelligence , 44(7):3523– 3542, 2021. 1

    2021

  47. [55]

    A dataset of multi-illumination images in the wild

    Lukas Murmann, Michael Gharbi, Miika Aittala, and Fredo Durand. A dataset of multi-illumination images in the wild. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4080–4089, 2019. 3

    2019

  48. [56]

    Difareli: Diffusion face relighting

    Puntawat Ponglertnapakorn, Nontawat Tritrong, and Supa- sorn Suwajanakorn. Difareli: Diffusion face relighting. In Proceedings of the IEEE/CVF international conference on computer vision, pages 22646–22657, 2023. 2

    2023

  49. [57]

    Deshadownet: A multi-context embedding deep network for shadow removal

    Liangqiong Qu, Jiandong Tian, Shengfeng He, Yandong Tang, and Rynson WH Lau. Deshadownet: A multi-context embedding deep network for shadow removal. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 4067–4075, 2017. 1, 3

    2017

  50. [58]

    Lite2relight: 3d-aware single image portrait relight- ing

    Pramod Rao, Gereon Fox, Abhimitra Meka, Mallikar- jun BR, Fangneng Zhan, Tim Weyrich, Bernd Bickel, Hanspeter Pfister, Wojciech Matusik, Mohamed Elgharib, et al. Lite2relight: 3d-aware single image portrait relight- ing. In ACM SIGGRAPH 2024 Conference Papers , pages 1–12, 2024. 1

    2024

  51. [59]

    Nerf for outdoor scene relighting

    Viktor Rudnev, Mohamed Elgharib, William Smith, Lingjie Liu, Vladislav Golyanik, and Christian Theobalt. Nerf for outdoor scene relighting. In European Conference on Com- puter Vision, pages 615–631. Springer, 2022. 2

    2022

  52. [60]

    Superglue: Learning feature matching with graph neural networks

    Paul-Edouard Sarlin, Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabinovich. Superglue: Learning feature matching with graph neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4938–4947, 2020. 1, 2

    2020

  53. [61]

    Image alignment and stitching

    Richard Szeliski and Richard Szeliski. Image alignment and stitching. Computer Vision: Algorithms and Applications , pages 401–441, 2022. 2

    2022

  54. [62]

    Rotation invariant iris recognition method adaptive to ambient lighting variation

    Hironobu Takano, Hiroki Kobayashi, and Kiyomi Naka- mura. Rotation invariant iris recognition method adaptive to ambient lighting variation. IEICE transactions on infor- mation and systems, 90(6):955–962, 2007. 2

    2007

  55. [63]

    Relight my nerf: A dataset for novel view synthe- sis and relighting of real world objects

    Marco Toschi, Riccardo De Matteo, Riccardo Spezialetti, Daniele De Gregorio, Luigi Di Stefano, and Samuele Salti. Relight my nerf: A dataset for novel view synthe- sis and relighting of real world objects. In Proceedings of the IEEE/CVF conference on computer vision and patter...

    2023

  56. [64]

    Pdc-net+: Enhanced probabilistic dense corre- spondence network

    Prune Truong, Martin Danelljan, Radu Timofte, and Luc Van Gool. Pdc-net+: Enhanced probabilistic dense corre- spondence network. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(8):10247–10266, 2023. 2

    2023

  57. [65]

    Wsrd: A novel benchmark for high resolution image shadow removal

    Florin-Alexandru Vasluianu, Tim Seizinger, and Radu Tim- ofte. Wsrd: A novel benchmark for high resolution image shadow removal. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 1826–1835, 2023. 1, 2, 3

    2023

  58. [66]

    Towards image ambient lighting normalization

    Florin-Alexandru Vasluianu, Tim Seizinger, Zongwei Wu, Rakesh Ranjan, and Radu Timofte. Towards image ambient lighting normalization. arXiv preprint arXiv:2403.18730 ,

    arXiv

  59. [67]

    Attention is all you need

    A Vaswani. Attention is all you need. Advances in Neural Information Processing Systems, 2017. 3

    2017

  60. [68]

    Interpretable object recognition by semantic prototype analysis

    Qiyang Wan, Ruiping Wang, and Xilin Chen. Interpretable object recognition by semantic prototype analysis. In Pro- ceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision, pages 800–809, 2024. 1

    2024

  61. [69]

    Stacked condi- tional generative adversarial networks for jointly learning shadow detection and shadow removal

    Jifeng Wang, Xiang Li, and Jian Yang. Stacked condi- tional generative adversarial networks for jointly learning shadow detection and shadow removal. In Proceedings of the IEEE conference on computer vision and pattern recog- nition, pages 1788–1797, 2018. 1, 2, 3

    2018

  62. [70]

    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. 3

    2024

  63. [71]

    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. 6

    2004

  64. [72]

    Uformer: A general u-shaped transformer for image restoration

    Zhendong Wang, Xiaodong Cun, Jianmin Bao, Wengang Zhou, Jianzhuang Liu, and Houqiang Li. Uformer: A general u-shaped transformer for image restoration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 17683–17693, 2022. 3, 6, 7

    2022

  65. [73]

    Deep retinex decomposition for low-light enhancement

    Chen Wei, Wenjing Wang, Wenhan Yang, and Jiaying Liu. Deep retinex decomposition for low-light enhancement. arXiv preprint arXiv:1808.04560, 2018. 3

    2018 arXiv

  66. [74]

    A novel automatic white balance method for digital still cam- eras

    Ching-Chih Weng, Homer Chen, and Chiou-Shann Fuh. A novel automatic white balance method for digital still cam- eras. In 2005 IEEE International Symposium on Circuits and Systems (ISCAS), pages 3801–3804. IEEE, 2005. 2

    2005

  67. [75]

    Cbam: Convolutional block attention module

    Sanghyun Woo, Jongchan Park, Joon-Young Lee, and In So Kweon. Cbam: Convolutional block attention module. In Proceedings of the European conference on computer vision (ECCV), pages 3–19, 2018. 6

    2018

  68. [76]

    Diffir: Efficient diffusion model for image restoration

    Bin Xia, Yulun Zhang, Shiyin Wang, Yitong Wang, Xing- long Wu, Yapeng Tian, Wenming Yang, and Luc Van Gool. Diffir: Efficient diffusion model for image restoration. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 13095–13105, 2023. 3

    2023

  69. [77]

    An efficient illumination normalization method for face recognition

    Xudong Xie and Kin-Man Lam. An efficient illumination normalization method for face recognition. Pattern Recogni- tion Letters, 27(6):609–617, 2006. 2

    2006

  70. [78]

    Yuen, and Ching Y

    Xiaohua Xie, Wei-Shi Zheng, Jianhuang Lai, Pong C. Yuen, and Ching Y . Suen. Normalization of face illumination based on large-and small-scale features. IEEE Transactions on Im- age Processing, 20(7):1807–1821, 2011. 2

    2011

  71. [79]

    Cretinex: A progressive color-shift aware retinex model for low-light im- age enhancement

    Han Xu, Hao Zhang, Xunpeng Yi, and Jiayi Ma. Cretinex: A progressive color-shift aware retinex model for low-light im- age enhancement. International Journal of Computer Vision, 132(9):3610–3632, 2024. 3

    2024

  72. [80]

    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

  73. [81]

    Generative portrait shadow removal

    Jae Shin Yoon, Zhixin Shu, Mengwei Ren, Cecilia Zhang, Yannick Hold-Geoffroy, Krishna Kumar Singh, and He Zhang. Generative portrait shadow removal. ACM Trans- actions on Graphics (TOG), 43(6):1–13, 2024. 1

    2024

  74. [82]

    Mambaout: Do we really need mamba for vision? arXiv preprint arXiv:2405.07992,

    Weihao Yu and Xinchao Wang. Mambaout: Do we really need mamba for vision? arXiv preprint arXiv:2405.07992,

    arXiv

  75. [83]

    Multi-stage progressive image restoration

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Multi-stage progressive image restoration. In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14821–14831, 2021. 3, 6, 7

    2021

  76. [84]

    Restormer: Efficient transformer for high-resolution image restoration

    Syed Waqas Zamir, Aditya Arora, Salman Khan, Mu- nawar Hayat, Fahad Shahbaz Khan, and Ming-Hsuan Yang. Restormer: Efficient transformer for high-resolution image restoration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5728–5739,

  77. [85]

    Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising

    Kai Zhang, Wangmeng Zuo, Yunjin Chen, Deyu Meng, and Lei Zhang. Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising. IEEE Transactions on Image Processing, 26(7):3142–3155, 2017. 3

    2017

  78. [86]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In CVPR, 2018. 6

    2018

  79. [87]

    Por- trait shadow manipulation

    Xuaner Zhang, Jonathan T Barron, Yun-Ta Tsai, Rohit Pandey, Xiuming Zhang, Ren Ng, and David E Jacobs. Por- trait shadow manipulation. ACM Transactions on Graphics (TOG), 39(4):78–1, 2020. 1

    2020

  80. [88]

    Zero-shot restoration of underexposed im- ages via robust retinex decomposition

    Anqi Zhu, Lin Zhang, Ying Shen, Yong Ma, Shengjie Zhao, and Yicong Zhou. Zero-shot restoration of underexposed im- ages via robust retinex decomposition. In 2020 IEEE Inter- national Conference on Multimedia and Expo (ICME), pages 1–6. IEEE, 2020. 3

    2020

  81. [89]

    Denoising dif- fusion models for plug-and-play image restoration

    Yuanzhi Zhu, Kai Zhang, Jingyun Liang, Jiezhang Cao, Bi- han Wen, Radu Timofte, and Luc Van Gool. Denoising dif- fusion models for plug-and-play image restoration. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1219–1229, 2023. 3

    2023

  82. [90]

    Oswald, and Marc Polle- feys

    Zihan Zhu, Songyou Peng, Viktor Larsson, Weiwei Xu, Hu- jun Bao, Zhaopeng Cui, Martin R. Oswald, and Marc Polle- feys. Nice-slam: Neural implicit scalable encoding for slam. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), pages 12...

    2022

  83. [2010]

    Springer Berlin Heidelberg. 2

  84. [2024]

    1, 2, 3, 4, 5, 6, 7, 8

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.