Pith. sign in

REVIEW 4 major objections 3 minor 41 references

Is-NeRF: In-scattering Neural Radiance Field for Blurred Images

T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Is-NeRF claims that modeling light as in-scattering lets NeRF recover sharp scenes from motion-blurred images.

desk verdict The abstract promises a deblur-NeRF with in-scattering, but the body is a networking paper; the submission is a mismatch and cannot be reviewed. read the letter →

arxiv 2508.13808 v1 pith:H2V5MAFU submitted 2025-08-19 cs.GR cs.CV

Nan Luo , Chenglin Ye , Jiaxu Li , Gang Liu , Bo Wan , Di Wang , Lupeng Liu , Jun Xiao This is my paper
classification cs.GRcs.CV
keywords neuralradiancefieldin-scatteringmotiondeblurringvolumerenderinglighttransportjointoptimizationimage
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

Is-NeRF sets out to prove that motion-blurred photographs, which current neural radiance field (NeRF) methods handle poorly because they assume light travels in straight lines from scene to camera, can instead be used to reconstruct sharp, geometrically accurate 3D scenes. The paper proposes replacing the straight-line volume renderer with a scattering-aware renderer that represents six light propagation phenomena through an in-scattering term, letting light paths bend and scatter into the camera ray. The network then jointly optimizes the NeRF scene parameters, the scattering parameters, and the camera motion on the same blurry images, so the blur is explained by the combination of scene and light transport rather than treated as noise. If the claim holds, this would make blurry real-world footage usable for high-fidelity novel-view synthesis and 3D reconstruction, going beyond prior deblur-NeRF methods that stick to straight-ray rendering.

What carries the argument

The key machinery is the in-scattering representation: a modified volume-rendering equation in which, at each sample along a camera ray, the radiance accumulates light scattered inward from directions off the ray, rather than only light emitted and absorbed on the ray itself. The paper folds six common light propagation phenomena into this representation, which is what allows the renderer to model complex light paths that a straight-line ray cannot represent. It does the work of turning motion blur from an artifact to be suppressed into a physical signal to be matched, and it is the component that, together with adaptive scattering-direction and sampling-interval selection, lets the joint op

What would settle it

Generate test images with a known blur source that the in-scattering representation cannot express—for example, defocus blur or a rolling-shutter warp—and check whether the recovered geometry departs from ground truth while the rendered blur still matches the input; alternatively, on data with known camera trajectories, start with deliberately broad scattering parameters and see whether the recovered geometry drifts even though the blur-fitting loss stays low.

Watch

Extended reading notes

Core claim

The central claim is that the straight-line volume-rendering equation used by every existing NeRF variant is too rigid for motion-blurred imagery, because blur corresponds to light that reaches a pixel from multiple directions and times along a path that does not lie on a single ray. Is-NeRF extends the volume renderer with an explicit in-scattering term that unifies six common light propagation phenomena, and uses this scattering-aware pipeline to accumulate radiance. On top of this, an adaptive learning strategy determines scattering directions and sampling intervals automatically, and the full network jointly optimizes NeRF parameters, scattering parameters, and camera motions. The paper

Load-bearing premise

The load-bearing premise is that real-world motion blur is faithfully captured by the six light-propagation phenomena folded into the in-scattering representation, and that jointly optimizing scene, scattering, and camera-motion parameters is identifiable—meaning a wrong sharp scene cannot be hidden behind a compensating scattering volume while still fitting the observed blur.

Editorial extensions

If this is right

  • Motion-blurred images become viable training data for NeRF, removing the need to pre-deblur or discard such captures.
  • The scattering-aware rendering pipeline could be carried over to other complex light-path settings, including haze, underwater scattering, and refractive distortion.
  • Jointly optimizing scene, scattering, and camera motion offers a path to disentangle scene geometry from light transport, potentially improving geometric accuracy in low-quality footage.
  • Adaptive direction and interval selection may cut sampling cost for fine details, making deblurred reconstruction faster.

Reading between the lines

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

  • The paper does not specify whether the six phenomena are exhaustive; a natural test is whether blur types like defocus or rolling-shutter warp, which are common in real footage, fall outside the in-scattering model and would need additional terms.
  • Identifiability is the hidden risk: if many (scene, scattering) combinations fit the same blur, the optimizer could trade true geometry for scattering effects. One way to test this is to remove scattering regularization and measure how much the recovered depth changes while the blur loss stays nearly constant.
  • The adaptive scattering-direction learning resembles a learned importance sampler over light paths; connecting it to existing ray-sampling strategies in NeRF might expose a common principle for handling non-straight ray transport.
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 / 3 minor

Summary. The submission presents the abstract of a paper titled "Is-NeRF: In-scattering Neural Radiance Field for Blurred Images," which claims a new scattering-aware volume rendering pipeline that unifies six light-propagation phenomena, adaptively determines scattering directions and sampling intervals, and jointly optimizes NeRF parameters, scattering parameters, and camera motions to recover sharp scenes from motion-blurred images. However, the provided full text is an unrelated manuscript titled "Reinforcement Learning-based Adaptive Path Selection for Programmable Networks," concerning Stochastic Learning Automata for in-network traffic steering in P4-programmable switches. The body contains no NeRF equations, no scattering model, no volume rendering derivation, no blur-formation model, and no experiments related to Is-NeRF. The central claims of the abstract are therefore entirely unsupported by the manuscript text.

Significance. If the claimed result were established, a scattering-aware NeRF that genuinely unifies six light-propagation phenomena and handles complex lightpaths for motion-blurred images would be a notable contribution to neural rendering and deblurring. The abstract promises novelty and state-of-the-art performance. However, the submission as provided does not allow any assessment of the method, the derivation, the experiments, or the reproducibility of the claims. There is no code, no proofs, no parameter-free derivations, and no falsifiable predictions to evaluate. The significance is contingent on content that is absent from the manuscript, so the work cannot currently be considered a valid contribution in its present form.

major comments (4)
  1. [Full text (entire body)] The body of the submitted manuscript is "Reinforcement Learning-based Adaptive Path Selection for Programmable Networks" (arXiv:2508.13806v2), not the Is-NeRF paper described in the abstract. The body contains no mention of NeRF, in-scattering, volume rendering, blur formation, scattering parameters, or camera motion. There is no derivation, no algorithm, and no experimental validation for any claim in the abstract. This is a load-bearing failure: the central assertion of the paper, including the six-phenomena unification and the joint optimization, is unsupported by any substantive content in the manuscript.
  2. [Abstract / Method] The abstract claims that Is-NeRF "unifies six common light propagation phenomena through an in-scattering representation" and establishes a "scattering-aware volume rendering pipeline." No definition of these six phenomena, the in-scattering representation, or the rendering equation appears anywhere in the provided text. Without this derivation, the claimed contribution cannot be checked, and the sentence about "adaptive learning strategy" for scattering directions and sampling intervals is likewise unverifiable.
  3. [Abstract / Evaluation] The abstract states "Comprehensive evaluations demonstrate that it effectively handles complex real-world scenarios, outperforming state-of-the-art approaches." The full text contains no experimental section, no datasets, no baselines, no metrics, and no results for Is-NeRF. This claim is formally unsupported. The only experiments present concern Mininet-based network telemetry, which are unrelated.
  4. [Identifiability / joint optimization] Even setting aside the mismatch, the abstract's joint optimization of NeRF parameters, scattering parameters, and camera-motion parameters on the same blurry images raises a serious identifiability risk: without additional constraints or a dedicated regularizer, the optimizer may trade off a wrong sharp scene against scattering parameters to explain the blur. Since the manuscript provides no formulation, no constraints, and no ablation, this risk is neither analyzed nor mitigated. The point is not that the idea is impossible, but that the submitted text gives no way to evaluate it.
minor comments (3)
  1. [Title / Abstract / Metadata] The title and abstract refer to Is-NeRF and arXiv number 2508.13808, while the full text carries the footer "arXiv:2508.13806v2" and its own abstract and references. The submission metadata is inconsistent, which prevents the reader from identifying the intended paper.
  2. [References] The reference list in the full text appears to support the networking paper only, with no citations to NeRF, volume rendering, scattering, or deblurring literature. The abstract references no prior NeRF works by name, making contextual positioning impossible.
  3. [Formatting] The full text is heavily truncated in the provided version (e.g., Section III ends mid-sentence and later sections are only enumerated), making even the unrelated networking paper incomplete. This further complicates any attempt to evaluate the submission.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity can be established: the claimed Is-NeRF derivation chain is entirely absent from the manuscript body, which contains an unrelated networking paper; the abstract's joint-optimization statement is an unsupported assertion, not a reduction to its inputs.

full rationale

The manuscript's abstract claims a scattering-aware NeRF pipeline, but the full text provided is a different paper, 'Reinforcement Learning-based Adaptive Path Selection for Programmable Networks' (arXiv:2508.13806v2), with no mention of NeRF, in-scattering, volume rendering, blur formation, or camera motion. There is therefore no derivation chain to walk. The only load-bearing statement is the abstract's promise that 'The proposed network jointly optimizes NeRF parameters, scattering parameters, and camera motions to recover fine-grained scene representations from blurry images.' This is a fitting-style claim, but no equations or evaluation are present to show that the recovered scene is defined by the fit; hence the circularity patterns (self-definition, fitted input called prediction, self-citation, uniqueness import, ansatz smuggling, renaming) cannot be exhibited. The absence of any supporting body text is a serious integrity/support problem—and under the reviewing rule this is explicitly flagged as missing support—but it is not circularity. No self-citations, uniqueness theorems, or ansatz-by-citation appear. The abstract's 'Comprehensive evaluations' are likewise unsupported by any experiments in the body. On the evidence available, the derivation reduces to nothing because no derivation exists, not because it reduces to its inputs.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The claimed method rests on the blur model, the rendering integral, and identifiability of the joint optimization, all asserted in the abstract and never formalized in the provided text. Three fitted parameter families are named in the abstract (scattering parameters, camera motions, adaptive sampling intervals). No new physical entities are introduced by the abstract; the in-scattering representation is a modeling abstraction, and the body text (an unrelated paper) does not bear on this ledger.

free parameters (3)
  • Scattering parameters (directions and coefficients)
    Jointly optimized with NeRF parameters to match blurry input images (abstract: 'jointly optimizes NeRF parameters, scattering parameters, and camera motions').
  • Camera motion parameters
    Jointly optimized per input image to deblur; fitted to the blurry views rather than derived from a physical model.
  • Sampling intervals
    Adaptively determined by the learning strategy, not derived from first principles (abstract: 'autonomous determining of scattering directions and sampling intervals').
assumptions (3)
  • domain assumption Motion blur in images is produced by light paths that the six unified in-scattering phenomena represent.
    Stated in the abstract as the premise for the scattering-aware pipeline.
  • domain assumption A scattering-aware volume rendering integral is the correct physical model for the camera's image formation under blur.
    Assumed by the proposed rendering pipeline; no derivation is present in the provided text.
  • domain assumption Joint optimization over NeRF, scattering, and camera-motion parameters is identifiable from blurry training views.
    Required for the method to recover the true sharp scene rather than any blur-explaining scene; not established anywhere in the submission.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Is-NeRF: In-scattering Neural Radiance Field for Blurred Images." pith.science (2026). https://pith.science/paper/H2V5MAFU

@misc{pith2026250813808,
  author       = {Pith},
  title        = {Pith review of: Is-NeRF: In-scattering Neural Radiance Field for Blurred Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H2V5MAFU}},
  note         = {Machine review of arXiv:2508.13808}
}
read the original abstract

Neural Radiance Fields (NeRF) has gained significant attention for its prominent implicit 3D representation and realistic novel view synthesis capabilities. Available works unexceptionally employ straight-line volume rendering, which struggles to handle sophisticated lightpath scenarios and introduces geometric ambiguities during training, particularly evident when processing motion-blurred images. To address these challenges, this work proposes a novel deblur neural radiance field, Is-NeRF, featuring explicit lightpath modeling in real-world environments. By unifying six common light propagation phenomena through an in-scattering representation, we establish a new scattering-aware volume rendering pipeline adaptable to complex lightpaths. Additionally, we introduce an adaptive learning strategy that enables autonomous determining of scattering directions and sampling intervals to capture finer object details. The proposed network jointly optimizes NeRF parameters, scattering parameters, and camera motions to recover fine-grained scene representations from blurry images. Comprehensive evaluations demonstrate that it effectively handles complex real-world scenarios, outperforming state-of-the-art approaches in generating high-fidelity images with accurate geometric details.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

41 extracted references · 38 canonical work pages

  1. [1]

    T.; Mildenhall, B.; Tancik, M.; Hedman, P.; Martin-Brualla, R.; and Srinivasan, P

    Barron, J. T.; Mildenhall, B.; Tancik, M.; Hedman, P.; Martin-Brualla, R.; and Srinivasan, P. P. 2021. Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 5855--5864

    work page 2021

  2. [2]

    T.; Mildenhall, B.; Verbin, D.; Srinivasan, P

    Barron, J. T.; Mildenhall, B.; Verbin, D.; Srinivasan, P. P.; and Hedman, P. 2022. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5470--5479

    work page 2022

  3. [3]

    Cao, A.; and Johnson, J. 2023. Hexplane: A fast representation for dynamic scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 130--141

    work page 2023

  4. [4]

    Chen, A.; Xu, Z.; Geiger, A.; Yu, J.; and Su, H. 2022. Tensorf: Tensorial radiance fields. In European conference on computer vision, 333--350. Springer

    work page 2022

  5. [5]

    Cui, Z.; Gu, L.; Sun, X.; Ma, X.; Qiao, Y.; and Harada, T. 2024. Aleth-NeRF: Illumination Adaptive NeRF with Concealing Field Assumption

    work page 2024

  6. [6]

    R.; Recht, B.; and Kanazawa, A

    Fridovich-Keil, S.; Meanti, G.; Warburg, F. R.; Recht, B.; and Kanazawa, A. 2023. K-planes: Explicit radiance fields in space, time, and appearance. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 12479--12488

    work page 2023

  7. [7]

    Fridovich-Keil, S.; Yu, A.; Tancik, M.; Chen, Q.; Recht, B.; and Kanazawa, A. 2022. Plenoxels: Radiance fields without neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 5501--5510

    work page 2022

  8. [8]

    Gao, C.; Sun, Q.; Zhu, J.; and Chen, J. 2025. MBS-NeRF: reconstruction of sharp neural radiance fields from motion-blurred sparse images. Scientific Reports, 15(1)

    work page 2025

Show all 41 references
  1. [9]

    Huang, X.; Zhang, Q.; Feng, Y.; Li, H.; Wang, X.; and Wang, Q. 2022. Hdr-nerf: High dynamic range neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 18398--18408

    2022

  2. [10]

    Kerbl, B.; Kopanas, G.; Leimk \"u hler, T.; and Drettakis, G. 2023. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42(4): 139--1

    2023

  3. [11]

    Kupyn, O.; Martyniuk, T.; Wu, J.; and Wang, Z. 2019. Deblurgan-v2: Deblurring (orders-of-magnitude) faster and better. In Proceedings of the IEEE/CVF international conference on computer vision, 8878--8887

    2019

  4. [12]

    Lee, B.; Lee, H.; Ali, U.; and Park, E. 2024 a . Sharp-NeRF: Grid-Based Fast Deblurring Neural Radiance Fields Using Sharpness Prior. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 3709--3718

    2024

  5. [13]

    Lee, B.; Lee, H.; Sun, X.; Ali, U.; and Park, E. 2024 b . Deblurring 3D Gaussian Splatting. arXiv:2401.00834

    2024 arXiv

  6. [14]

    Lee, D.; Lee, M.; Shin, C.; and Lee, S. 2023. Dp-nerf: Deblurred neural radiance field with physical scene priors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 12386--12396

    2023

  7. [15]

    Levy, D.; Peleg, A.; Pearl, N.; Rosenbaum, D.; Akkaynak, D.; Korman, S.; and Treibitz, T. 2023. Seathru-nerf: Neural radiance fields in scattering media. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 56--65

    2023

  8. [16]

    Li, T.; Slavcheva, M.; Zollhoefer, M.; Green, S.; Lassner, C.; Kim, C.; Schmidt, T.; Lovegrove, S.; Goesele, M.; Newcombe, R.; et al. 2022. Neural 3d video synthesis from multi-view video. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5521--5531

    2022

  9. [17]

    Li, Z.; Niklaus, S.; Snavely, N.; and Wang, O. 2021. Neural scene flow fields for space-time view synthesis of dynamic scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 6498--6508

    2021

  10. [18]

    F.; and Lee, G

    Low, W. F.; and Lee, G. H. 2025. Deblur e-NeRF: NeRF fromMotion-Blurred Events under High-speed orLow-light Conditions. In European Conference on Computer Vision

    2025

  11. [19]

    S.; Song, X.; Yu, H.; Lin, Z.; and Peng, L

    Luthra, A.; Gantha, S. S.; Song, X.; Yu, H.; Lin, Z.; and Peng, L. 2024. Deblur-NSFF: Neural Scene Flow Fields for Blurry Dynamic Scenes. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 3658--3667

    2024

  12. [20]

    Ma, L.; Li, X.; Liao, J.; Zhang, Q.; Wang, X.; Wang, J.; and Sander, P. V. 2022. Deblur-nerf: Neural radiance fields from blurry images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 12861--12870

    2022

  13. [21]

    S.; Barron, J

    Martin-Brualla, R.; Radwan, N.; Sajjadi, M. S.; Barron, J. T.; Dosovitskiy, A.; and Duckworth, D. 2021. Nerf in the wild: Neural radiance fields for unconstrained photo collections. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7210--7219

    2021

  14. [22]

    P.; Tancik, M.; Barron, J

    Mildenhall, B.; Srinivasan, P. P.; Tancik, M.; Barron, J. T.; Ramamoorthi, R.; and Ng, R. 2021. Nerf: Representing scenes as neural radiance fields for view synthesis. Communications of the ACM, 65(1): 99--106

    2021

  15. [23]

    M \"u ller, T.; Evans, A.; Schied, C.; and Keller, A. 2022. Instant neural graphics primitives with a multiresolution hash encoding. ACM transactions on graphics (TOG), 41(4): 1--15

    2022

  16. [24]

    Niemeyer, M.; and Geiger, A. 2021. Giraffe: Representing scenes as compositional generative neural feature fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 11453--11464

    2021

  17. [25]

    Park, H.; and Mu Lee, K. 2017. Joint estimation of camera pose, depth, deblurring, and super-resolution from a blurred image sequence. In Proceedings of the IEEE International Conference on Computer Vision, 4613--4621

    2017

  18. [26]

    Peng, S.; Dong, J.; Wang, Q.; Zhang, S.; Shuai, Q.; Zhou, X.; and Bao, H. 2021. Animatable neural radiance fields for modeling dynamic human bodies. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 14314--14323

    2021

  19. [27]

    Philip, J.; Morgenthaler, S.; Gharbi, M.; and Drettakis, G. 2021. Free-viewpoint indoor neural relighting from multi-view stereo. ACM Transactions on Graphics (TOG), 40(5): 1--18

    2021

  20. [28]

    Qi, Y.; Zhu, L.; Zhang, Y.; and Li, J. 2023. E2NeRF: Event Enhanced Neural Radiance Fields from Blurry Images. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 13254--13264

    2023

  21. [29]

    Qi, Y.; Zhu, L.; Zhao, Y.; Bao, N.; and Li, J. 2024. Deblurring Neural Radiance Fields with Event-driven Bundle Adjustment. In Proceedings of the 32nd ACM International Conference on Multimedia, MM '24, 9262–9270. New York, NY, USA: Association for Computing Machinery. ISBN 97...

    2024

  22. [30]

    Ramazzina, A.; Bijelic, M.; Walz, S.; Sanvito, A.; Scheuble, D.; and Heide, F. 2023. ScatterNeRF: Seeing Through Fog with Physically-Based Inverse Neural Rendering. arXiv:2305.02103

    2023 arXiv

  23. [31]

    Schwarz, K.; Liao, Y.; Niemeyer, M.; and Geiger, A. 2020. Graf: Generative radiance fields for 3d-aware image synthesis. Advances in Neural Information Processing Systems, 33: 20154--20166

    2020

  24. [32]

    Son, H.; Lee, J.; Lee, J.; Cho, S.; and Lee, S. 2021. Recurrent video deblurring with blur-invariant motion estimation and pixel volumes. ACM Transactions on Graphics (TOG), 40(5): 1--18

    2021

  25. [33]

    P.; Deng, B.; Zhang, X.; Tancik, M.; Mildenhall, B.; and Barron, J

    Srinivasan, P. P.; Deng, B.; Zhang, X.; Tancik, M.; Mildenhall, B.; and Barron, J. T. 2021. Nerv: Neural reflectance and visibility fields for relighting and view synthesis. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7495--7504

    2021

  26. [34]

    Su, S.; Delbracio, M.; Wang, J.; Sapiro, G.; Heidrich, W.; and Wang, O. 2017. Deep video deblurring for hand-held cameras. In Proceedings of the IEEE conference on computer vision and pattern recognition, 1279--1288

    2017

  27. [35]

    Su, S.-Y.; Yu, F.; Zollh \"o fer, M.; and Rhodin, H. 2021. A-nerf: Articulated neural radiance fields for learning human shape, appearance, and pose. Advances in Neural Information Processing Systems, 34: 12278--12291

    2021

  28. [36]

    Sun, J.; Xie, Y.; Chen, L.; Zhou, X.; and Bao, H. 2021. Neuralrecon: Real-time coherent 3d reconstruction from monocular video. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 15598--15607

    2021

  29. [37]

    Tao, X.; Gao, H.; Shen, X.; Wang, J.; and Jia, J. 2018. Scale-recurrent network for deep image deblurring. In Proceedings of the IEEE conference on computer vision and pattern recognition, 8174--8182

    2018

  30. [38]

    Wang, D.; Wang, Q.; He, L.; Gao, X.; and Tian, Y. 2020. Joint and individual matrix factorization hashing for large-scale cross-modal retrieval. Pattern recognition, 107: 107479

    2020

  31. [39]

    Wang, P.; Zhao, L.; Ma, R.; and Liu, P. 2023. Bad-nerf: Bundle adjusted deblur neural radiance fields. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 4170--4179

    2023

  32. [40]

    Wang, X.; Yin, Z.; Zhang, F.; Feng, D.; and Wang, Z. 2024. MP-NeRF: More refined deblurred neural radiance field for 3D reconstruction of blurred images. Knowledge-based systems, 290(Apr.22): 1.1--1.10

    2024

  33. [41]

    W.; Arora, A.; Khan, S.; Hayat, M.; Khan, F

    Zamir, S. W.; Arora, A.; Khan, S.; Hayat, M.; Khan, F. S.; Yang, M.-H.; and Shao, L. 2021. Multi-stage progressive image restoration. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 14821--14831

    2021

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.