Pith. sign in

REVIEW 3 major objections 7 minor 1 cited by

Towards Degradation-Robust Reconstruction in Generalizable NeRF

T0 review · 3 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Generalizable NeRFs can be made robust to blur, noise, and adversarial corruption by replacing the image-feature extractor with a depth-aware module.

desk verdict A useful dataset and a simple, consistently effective plug-in module for degradation-robust GNeRF; the depth-supervision loop is under-validated but fixable. read the letter →

arxiv 2411.11691 v1 pith:PJ32EDMO submitted 2024-11-18 cs.CV

Chan Ho Park , Ka Leong Cheng , Zhicheng Wang , Qifeng Chen This is my paper · ORCID
classification cs.CV
keywords generalizableneuralradiancefieldsdegradationrobustnessmotionblurdatasetdepthestimationdifferentiablewarpingnovelviewsynthesisimagerestorationself-supervised
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

The paper argues that generalizable neural radiance fields (GNeRFs), which synthesize new views from a handful of source images without per-scene training, lose accuracy when those source images are blurry, noisy, or adversarially perturbed. It proposes a model-agnostic replacement for the standard image-feature extractor: first estimate depth from nearby degraded views, warp those views onto each source view using the predicted depth, then restore a clean latent representation with an auxiliary denoiser. On three popular GNeRF designs with different aggregation mechanisms, the module consistently improves novel-view PSNR, SSIM, and LPIPS under blur, noise, adversarial attacks, and blur-plus-noise, and it also steadies depth prediction as blur increases. The paper also introduces the Objaverse Blur Dataset, over 50,000 images rendered from hundreds of 3D models along simulated camera trajectories, as a large-scale benchmark for training and evaluating degradation-robust GNeRFs.

What carries the argument

The central object is the 3D-degradation-aware feature extractor $\mathcal{F}$, a two-stage module consisting of a self-supervised depth estimator $\mathcal{D}$ and an auxiliary restoration head $\mathcal{R}$. The load-bearing operation is the differentiable homography warping of Eq. (5), which uses estimated depth and relative camera pose to align nearby views to each source view; the concatenated aligned stack $J_i$ of Eq. (6) is what lets the restoration head exploit multi-view, depth-consistent evidence. The depth loss of Eq. (7) closes a self-supervised loop from the GNeRF's own fine depth prediction, and the restoration loss of Eq. (8) provides clean-image supervision; together with the photometric loss they form the total loss of Eq. (9). Because the module outputs a feature tensor identical in shape to the original CNN feature map, it plugs into any GNeRF without changing the volume-rendering stage.

What would settle it

Train the module with an independent depth signal, such as depths estimated from clean views of the same training scenes, instead of the GNeRF's fine depth as pseudo-ground truth. If the robustness gains on blurry or noisy test views disappear, or if the estimated depths drift from the independent depths, the self-referential depth loop is carrying the result.

Watch

Extended reading notes

Core claim

The paper's central claim is that the degradation vulnerability of GNeRF models is largely a geometry problem: when source images are degraded, the features sampled along a ray become inconsistent, depth prediction drifts, and rendering quality collapses. To counter this, the proposed 3D-degradation-aware feature extractor first estimates a coarse depth map for each source view from its nearest neighboring views using a multi-scale patch-match stereo cost volume, then applies differentiable homography warping, $\pi^k_i(p) = K_k \cdot (R^k_i \cdot (K_i^{-1} \cdot p \cdot D_i[p]) + t^k_i)$, to align the nearby views and their depths onto the source view. The aligned stack $J_i = \text{Concat}(I_i, D_i, \tilde{I}_1, \ldots, \tilde{I}_K, \tilde{D}_1, \ldots, \tilde{D}_K)$ is fed to a restoration head built from multi-Dconv head transposed attention blocks, whose output features have the same shape as a conventional CNN feature map, so any existing GNeRF can use them without altering its renderer. The depth estimator is trained without ground-truth depth by regressing the GNeRF's own fine depth prediction through a self-supervised loss, $\mathcal{L}_{\mathrm{depth}} = \text{SmoothL1}(\hat{D}_{\mathrm{tar}}, D_{\mathrm{tar}})$, while the restoration head is supervised by clean source images through an auxiliary restoration loss. Combined with the original photometric loss, these terms train the module end-to-end along with the GNeRF; the paper reports consistent quantitative and visual improvements across blur levels, noise gain levels, adversarial perturbations, and combined blur-plus-noise degradation.

Load-bearing premise

The depth estimator is trained to imitate the same GNeRF depth prediction it helps produce, so if that prediction is already corrupted by degradation, the module may learn to lock in wrong geometry instead of correcting it.

Editorial extensions

If this is right

  • The robustness gain transfers across GNeRF architectures, including attention-based, image-based, and cost-volume-based models, so the mechanism is not tied to one rendering formula.
  • Stronger degradation sees larger relative gains: noise experiments show improvements up to +0.63 PSNR at gain 20, adversarial evaluation gains reach +1.22 PSNR, and blur gains remain positive across all four blur levels.
  • Depth prediction becomes more stable: the absolute and relative deviation of predicted depth from the clean-image prediction drops by roughly 28 to 35 percent when the module is added under blur.
  • The module also improves rendering on clean source images, indicating that the 3D-aligned features help representation quality beyond robustness alone.
  • More source views amplify the benefit, which suggests the alignment mechanism increasingly exploits multi-view redundancy as the number of degraded inputs grows.

Reading between the lines

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

  • The self-supervised depth loop of Eq. (7) is the most fragile link; swapping the GNeRF's fine depth for an externally validated depth map, or masking the loss where the GNeRF depth is uncertain, could push gains beyond what the paper reports.
  • Because the new dataset renders blur by averaging frames along a camera trajectory, the degradation is depth-dependent; a direct comparison against image-space uniform blur on the same views would test whether the geometry-alignment mechanism is truly what carries the improvement.
  • The same aligned-feature design could extend to other structured degradations such as haze, rain, or low light by exchanging the restoration head, since the warping stage only requires poses and estimated depth.
  • If the module's benefit grows with view count, it could combine naturally with inference-time selection of additional source views, an extension the paper leaves unexplored.
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

3 major / 7 minor

Summary. This paper studies the robustness of generalizable NeRF (GNeRF) models when source images are degraded by blur, noise, adversarial perturbations, or blur combined with noise, and makes two contributions. First, it introduces the Objaverse Blur Dataset, a Blender-rendered dataset of over 50,000 motion-blurred images generated from 250 Objaverse models at four blur levels via an algorithm that averages latent images along sampled camera trajectories for 3D-consistent degradation. Second, it proposes a model-agnostic 3D-degradation-aware feature extractor: a self-supervised depth estimator (PatchMatchNet-style), a differentiable depth-based warping that aligns nearby views (Eq. 5), and a Restormer-based restoration head producing degradation-robust features for the downstream GNeRF aggregator. Training adds two auxiliary losses — depth pseudo-supervision against the GNeRF's own fine depth prediction (Eq. 7) and a restoration loss against clean source images (Eq. 8). Experiments with NAN, GeoNeRF, and GNT report mostly consistent PSNR/SSIM/LPIPS gains across four degradation settings, a depth-stability analysis (Table 5), and an inference-speed comparison (Table 6).

Significance. The empirical pattern is the paper's main strength: the module improves three structurally different GNeRF backbones (image-based NAN, cost-volume-based GeoNeRF, attention-based GNT) across blur, noise, adversarial, and combined blur-and-noise degradations, and the main metrics are evaluated on held-out target views, so the reported reconstruction gains are not fitted quantities. The dataset construction is specified in enough detail (Algorithm 1, camera and trajectory sampling, scene-dependent blur weights) to be reproducible, and the paper honestly concedes in Appendix E that the method may fail under larger degradation or inaccurate poses, consistent with the diminishing gains at high blur/noise levels in Table 4. If these results hold, the dataset and the plug-in module are practically useful resources for the GNeRF community. The principal weakness is that the stated mechanism — robustness through geometry stabilization via depth estimation and warping — is not directly validated: depth is only reported for stability against the clean-input prediction, never for accuracy, and no ablation isolates the depth/warping component from the restoration head.

major comments (3)
  1. [§4.3, Eq. (7); §6.6, Table 5] The depth estimator D is supervised only by the GNeRF's own fine depth prediction D_hat_tar (Eq. 7), while the features consumed by that same GNeRF are produced by a warping that depends on D's output (Eq. 5); this self-referential loop is never validated against ground-truth depth, although the Objaverse Blur Dataset is rendered from 3D models and ground-truth depth is available at essentially zero cost. Table 5 measures only the deviation of predicted depth from the clean-input prediction (stability), not accuracy, so the paper's central mechanistic claim — that the robustness gain stems from correcting inaccurate geometry (Sec. 4) — remains unsupported. The photometric and restoration losses may anchor the loop, but this is not demonstrated. Please (i) report depth accuracy (absolute and relative error against rendered ground-truth depth) for D and for the GNeRF's fine depth with and without the module on the Objaverse Blur Dataset, and (ii) add an ablation that isolates the depth/warping path from the restoration head, for example by feeding oracle ground-truth depth to the warping or by removing the warping and keeping only the restoration head.
  2. [§6.1; Appendix D] The evaluation protocol on the newly introduced dataset is under-specified. The paper states that 52 settings were randomly chosen and that results are averaged over 13 scenes (Appendix D), but it does not state whether these test 3D models are held out from the training set, nor does it state the training protocol for the blur experiments — that is, whether NAN and GeoNeRF were trained from scratch on the Objaverse Blur Dataset, pre-trained elsewhere and fine-tuned, or evaluated with only the proposed module trained. Without this information the reported improvements cannot be interpreted as generalization to unseen scenes. Please specify the exact train/test model split, the number of training settings, and the training protocol, and reconcile the dataset size claims (over 1000 settings, 50,000 images) with the rendering recipe in Sec. 5, since 250 models times 4 levels times 34 images per viewpoint would require roughly one to two viewpoints per model to reach 50,000 images.
  3. [§6.3–6.4, Tables 2, 4, 10] The claim of consistent improvement is made without any measure of variance. Several deltas are very small — +0.05 dB at gain 4 in Table 2 and +0.08 dB at blur-4/gain-4 in Table 4 — and the per-scene adversarial results in Table 10 include negative deltas (Fortress −0.41 in the Adversarial-Clean block, Orchids −0.19 and −0.10 in the adversarial blocks), which contradicts the unqualified statement in Sec. 6.3 that the module is effective across all training and test combinations. Please report multiple seeds or per-setting standard deviations, and either qualify the adversarial claim or explain the negative per-scene cases.
minor comments (7)
  1. [Table 1 vs. Table 8] GeoNeRF* at blur level 3 is 23.78 dB in Table 1 but 23.79 dB in Table 8; duplicated numbers should be consistent across the main text and appendix.
  2. [§6.1 and Eq. (9)] The loss weights are called λdepth and λrec in the text but λ1 and λ2 in Eq. (9), and the sentence setting them to 1.0 and 0.01 does not say which weight is which; please clarify the mapping.
  3. [§6.2] The text refers to 'Tabs. 8, 9, and 10' when summarizing results, but the main-text tables are numbered 1–6 and the cited tables appear only in the appendix; cross-references should be consistent.
  4. [§4.3 and §4.1] Ldepth is defined only for the target view, while §4.1 states that depths are estimated for all N source images; please state how source-view depth estimates are supervised, if at all.
  5. [Eq. (8)] The summation index i = 1..K is unclear because the restoration head is described as operating on a single source view; please clarify the indexing and define I^GT_i as the clean counterpart of the i-th source image.
  6. [Fig. 2 and Eq. (2)] The figure caption uses the symbol S^{src} while Eq. (2) defines the feature set S_x = {s_i}; please unify the notation.
  7. [Sec. 2 (Related Work)] The relationship between the Objaverse Blur Dataset and the 3D-consistent corruption benchmark of [30] is mentioned only in passing; a sentence clarifying the difference (single corruption type, multi-level blur, GNeRF training scale) would strengthen the dataset positioning.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor non-load-bearing self-referential depth pseudo-supervision loop; main held-out reconstruction results are independent.

  1. other [Section 4.3, Eq. (7); Section 6.6, Table 5]
    "As we assume the absence of ground-truth depth, the depth estimator is supervised by the target view fine depth prediction ^Dtar of the GNeRF as the pseudo ground truth. ... Ldepth = SmoothL1( ^Dtar, Dtar). (7) ... to evaluate the stability of depth prediction over varying blur degradation levels, we have measured the change in absolute and relative depth prediction per pixel. ... the result also indicates that incorporating our module results in less deviation from the clean source image-based prediction."

    The depth estimator D is trained to match the GNeRF's own fine depth prediction ^Dtar (Eq. 7). D's predicted depth defines the differentiable warping (Eq. 5) and the aligned feature J_i (Eq. 6), which are the inputs from which the GNeRF produces ^Dtar. Thus the depth estimator's supervision signal is generated by a model that is itself conditioned on the depth estimator's output, forming a closed self-referential loop. Table 5 then reports this loop's 'stability' as deviation from the clean-source prediction, which is a self-consistency measure rather than an accuracy check against ground-truth depth (which the Blender-based dataset could have provided).

full rationale

The paper's central evaluation is self-contained: Tables 1-4 and 7-11 report PSNR/SSIM/LPIPS on held-out novel views against ground-truth rendered images, so these numbers are not fitted quantities. The proposed module is compared against strong baselines (original GNeRFs, Restormer pre-processing) under identical training data. The only self-referential element is the depth pseudo-supervision in Eq. (7), where the depth estimator is trained to reproduce the GNeRF's own fine depth prediction, which in turn depends on the depth estimator through warping (Eqs. 5-6). This is a genuine bootstrap loop, but it is not load-bearing for the main held-out reconstruction metrics, which are anchored by photometric and restoration losses against ground-truth clean images (Eqs. 8-9). The depth-stability analysis of Table 5 is a self-consistency measure rather than an accuracy measure, and the paper's Appendix E concedes failure under larger degradation; these are correctness/evidence concerns rather than demonstrations that a prediction equals its input by construction. No load-bearing self-citation or imported uniqueness theorem appears: NeRFool is an external citation used for motivation, and the GNeRF baselines are external methods. Overall circularity burden is low; score 2 reflects the minor non-load-bearing self-referential depth loop.

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

The central claim rests on a small set of hand-set hyperparameters, the availability of accurate poses, and the assumption that the GNeRF's own depth is a useful pseudo ground truth. The dataset generation also assumes that averaged latent renders are a good proxy for real motion blur.

free parameters (5)
  • lambda_1 (depth loss weight) = 1.0
    Chosen empirically to balance L_depth in Eq. (9).
  • lambda_2 (restoration loss weight) = 0.01
    Chosen empirically; annealed with alpha=0.99997 down to 0.01, as described in Sec. 6.1.
  • alpha (annealing rate) = 0.99997
    Set by hand to control the decay of L_restore weight.
  • number of averaged latent frames (m) = 34
    Dataset rendering parameter that sets blur strength; also n=34 frames per viewpoint.
  • camera trajectory delta_0 = 2.5
    Selected after tuning to cover reasonable blur ranges in Objaverse models (Sec. 5).
assumptions (5)
  • domain assumption The GNeRF's fine depth prediction is a reliable pseudo ground truth for training the depth estimator under degraded inputs.
    Invoked in Eq. (7) and Sec. 4.3; no independent depth supervision is used, and reliability under degradation is not analyzed.
  • domain assumption Accurate camera poses are available for all source images during training and inference.
    Used in Eq. (4) and the warping in Eq. (5); the authors list this as a limitation in Appendix E, noting COLMAP may be inaccurate under degradation.
  • domain assumption Motion blur can be approximated by averaging m latent images rendered along a sampled camera trajectory.
    Dataset generation in Algorithm 1; the approximation is validated only by one qualitative real-noise example, not by quantitative real-blur evaluation.
  • domain assumption Warping nearby views with the estimated depth improves the feature representation for restoration and reconstruction.
    The central design of Sec. 4.1 and 4.2; if the depth is wrong, the stacked aligned features may harm rather than help.
  • domain assumption The restoration head can recover a clean image from the depth-aligned stack and the learned features carry enough signal for NeRF volume rendering.
    Assumed by the auxiliary restoration loss in Eq. (8) and the feature replacement in Sec. 4.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Degradation-Robust Reconstruction in Generalizable NeRF." pith.science (2026). https://pith.science/paper/PJ32EDMO

@misc{pith2026241111691,
  author       = {Pith},
  title        = {Pith review of: Towards Degradation-Robust Reconstruction in Generalizable NeRF},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PJ32EDMO}},
  note         = {Machine review of arXiv:2411.11691}
}
read the original abstract

Generalizable Neural Radiance Field (GNeRF) across scenes has been proven to be an effective way to avoid per-scene optimization by representing a scene with deep image features of source images. However, despite its potential for real-world applications, there has been limited research on the robustness of GNeRFs to different types of degradation present in the source images. The lack of such research is primarily attributed to the absence of a large-scale dataset fit for training a degradation-robust generalizable NeRF model. To address this gap and facilitate investigations into the degradation robustness of 3D reconstruction tasks, we construct the Objaverse Blur Dataset, comprising 50,000 images from over 1000 settings featuring multiple levels of blur degradation. In addition, we design a simple and model-agnostic module for enhancing the degradation robustness of GNeRFs. Specifically, by extracting 3D-aware features through a lightweight depth estimator and denoiser, the proposed module shows improvement on different popular methods in GNeRFs in terms of both quantitative and visual quality over varying degradation types and levels. Our dataset and code will be made publicly available.

Figures

Figures reproduced from arXiv: 2411.11691 by the authors.

Figure 1
Figure 1. Visual comparison of the different degradation settings [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Training pipeline for degradation-robust GNeRF. This diagram illustrates the process for training a GNeRF that is robust to [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Examples of Objaverse Blur Dataset at different blur levels. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison of GNeRFs [28, 49, 53] on different degradations and datasets [45, 49]. Each row contains the nearest degraded source image, rendered RGB with and without our proposed module, and ground-truth target view. fairness, we train Restormer with the gi…
Figure 5
Figure 5. Figure 5: Qualitative evaluation on real low-light noise captured [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 7
Figure 7. Figure 7: Histogram of blur weights at different levels of [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Visual results of 3D Differentiable Warping through Depth Estimator [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Visual results of 3D-aware Auxiliary Restoration Head [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. R3eVision: A Survey on Robust Rendering, Restoration, and Enhancement for 3D Low-Level Vision

    cs.CV 2025-06 conditional novelty 6.0 of 10

    The survey formalizes degradation-aware rendering for 3D Low-Level Vision and organizes roughly 100 methods on super-resolution, deblurring, weather removal, restoration, and enhancement in NeRF and 3DGS pipelines.

Reference graph

Works this paper leans on

70 extracted references · 66 canonical work pages · cited by 1 Pith paper

  1. [1]

    G. Bhat, M. Danelljan, F. Yu, Luc Van Gool, and R. Timofte. Deep reparametrization of multi-frame super-resolution and denoising. In Proceedings of ICCV, pages 2440–2450. IEEE,

    work page

  2. [2]

    Dehazenet: An end-to-end system for sin- gle image haze removal

    Bolun Cai, Xiangmin Xu, Kui Jia, Chunmei Qing, and Dacheng Tao. Dehazenet: An end-to-end system for sin- gle image haze removal. IEEE Trans. Image Process., 25(11): 5187–5198, 2016. 2

    work page 2016

  3. [3]

    Hexplane: A fast representation for dynamic scenes

    Ang Cao and Justin Johnson. Hexplane: A fast representation for dynamic scenes. In Proceedings of CVPR, pages 130–141. IEEE, 2023. 1

    work page 2023

  4. [4]

    Mvsnerf: Fast general- izable radiance field reconstruction from multi-view stereo

    Anpei Chen, Zexiang Xu, Fuqiang Zhao, Xiaoshuai Zhang, Fanbo Xiang, Jingyi Yu, and Hao Su. Mvsnerf: Fast general- izable radiance field reconstruction from multi-view stereo. In Proceedings of ICCV, pages 14104–14113. IEEE, 2021. 1, 2, 6

    work page 2021

  5. [5]

    Pre-trained image processing transformer

    Hanting Chen, Yunhe Wang, Tianyu Guo, Chang Xu, Yiping Deng, Zhenhua Liu, Siwei Ma, Chunjing Xu, Chao Xu, and Wen Gao. Pre-trained image processing transformer. In Proceedings of CVPR, pages 12299–12310. Computer Vision Foundation / IEEE, 2021. 2

    work page

  6. [6]

    Simple baselines for image restoration

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration. In Proceedings of ECCV, pages 17–33. Springer, 2022. 2

    work page 2022

  7. [7]

    DehazeNeRF: Multiple Image Haze Removal and 3D Shape Reconstruction using Neural Radiance Fields

    Wei-Ting Chen, Yifan Wang, Sy-Yen Kuo, and Gordon Wet- zstein. Dehazenerf: Multiple image haze removal and 3d shape reconstruction using neural radiance fields. CoRR, abs/2303.11364, 2023. 1

    work page Pith review arXiv 2023

  8. [8]

    Hallucinated neural radiance fields in the wild

    Xingyu Chen, Qi Zhang, Xiaoyu Li, Yue Chen, Ying Feng, Xuan Wang, and Jue Wang. Hallucinated neural radiance fields in the wild. In Proceedings of CVPR, pages 12933– 12942. IEEE, 2022. 2

    work page 2022

Show all 70 references
  1. [9]

    Activating more pixels in image super-resolution 9 transformer

    Xiangyu Chen, Xintao Wang, Jiantao Zhou, Yu Qiao, and Chao Dong. Activating more pixels in image super-resolution 9 transformer. In Proceedings of CVPR, pages 22367–22377. IEEE, 2023. 2

    2023

  2. [10]

    Blender - a 3D modelling and rendering package

    Blender Online Community. Blender - a 3D modelling and rendering package. Blender Foundation, Stichting Blender Foundation, Amsterdam, 2018. 5

    2018

  3. [11]

    Aleth-nerf: Low-light condition view synthesis with conceal- ing fields

    Ziteng Cui, Lin Gu, Xiao Sun, Yu Qiao, and Tatsuya Harada. Aleth-nerf: Low-light condition view synthesis with conceal- ing fields. CoRR, abs/2303.05807, 2023. 2

    2023 arXiv

  4. [12]

    Objaverse: A universe of annotated 3d objects

    Matt Deitke, Dustin Schwenk, Jordi Salvador, Luca Weihs, Oscar Michel, Eli VanderBilt, Ludwig Schmidt, Kiana Ehsani, Aniruddha Kembhavi, and Ali Farhadi. Objaverse: A universe of annotated 3d objects. In Proceedings of CVPR , pages 13142–13153. IEEE, 2023. 2, 5, 6

    2023

  5. [13]

    Tenen- baum, and Jiajun Wu

    Yilun Du, Yinan Zhang, Hong-Xing Yu, Joshua B. Tenen- baum, and Jiajun Wu. Neural radiance flow for 4d view synthesis and video processing. In Proceedings of ICCV , pages 14304–14314. IEEE, 2021. 2

    2021

  6. [14]

    Burstormer: Burst image restoration and enhancement transformer

    Akshay Dudhane, Syed Waqas Zamir, Salman Khan, Fa- had Shahbaz Khan, and Ming-Hsuan Yang. Burstormer: Burst image restoration and enhancement transformer. In Proceed- ings of CVPR, pages 5703–5712. IEEE, 2023. 2, 4

    2023

  7. [15]

    Plenoxels: Radiance fields without neural networks

    Sara Fridovich-Keil, Alex Yu, Matthew Tancik, Qinhong Chen, Benjamin Recht, and Angjoo Kanazawa. Plenoxels: Radiance fields without neural networks. In Proceedings of CVPR, pages 5491–5500. IEEE, 2022. 1

    2022

  8. [16]

    Nerfool: Uncovering the vulnerability of generalizable neural radiance fields against adversarial perturbations

    Yonggan Fu, Ye Yuan, Souvik Kundu, Shang Wu, Shunyao Zhang, and Yingyan Celine Lin. Nerfool: Uncovering the vulnerability of generalizable neural radiance fields against adversarial perturbations. In Proceedings of ICML, 2023. 2, 3, 4, 6, 7, 8, 13

    2023

  9. [17]

    ”double- dip”: Unsupervised image decomposition via coupled deep- image-priors

    Yossi Gandelsman, Assaf Shocher, and Michal Irani. ”double- dip”: Unsupervised image decomposition via coupled deep- image-priors. In Proceedings of CVPR, pages 11026–11035. Computer Vision Foundation / IEEE, 2019. 4

    2019

  10. [18]

    Garbin, Marek Kowalski, Matthew Johnson, Jamie Shotton, and Julien P

    Stephan J. Garbin, Marek Kowalski, Matthew Johnson, Jamie Shotton, and Julien P. C. Valentin. Fastnerf: High-fidelity neural rendering at 200fps. In Proceedings of ICCV, pages 14326–14335. IEEE, 2021. 1

    2021

  11. [19]

    Girshick

    Ross B. Girshick. Fast R-CNN. In Proceedings of ICCV, pages 1440–1448. IEEE Computer Society, 2015. 5

    2015

  12. [20]

    Goodfellow, Jonathon Shlens, and Christian Szegedy

    Ian J. Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial examples. In Proceed- ings of ICLR, 2015. 2

    2015

  13. [21]

    Cascade cost volume for high-resolution multi-view stereo and stereo matching

    Xiaodong Gu, Zhiwen Fan, Siyu Zhu, Zuozhuo Dai, Feitong Tan, and Ping Tan. Cascade cost volume for high-resolution multi-view stereo and stereo matching. In Proceedings of CVPR, pages 2492–2501. Computer Vision Foundation / IEEE, 2020. 4

    2020

  14. [22]

    Deep decoder: Concise image representations from untrained non-convolutional net- works

    Reinhard Heckel and Paul Hand. Deep decoder: Concise image representations from untrained non-convolutional net- works. In Proceedings of ICLR. OpenReview.net, 2019. 4

    2019

  15. [23]

    Srinivasan, Ben Mildenhall, Jonathan T

    Peter Hedman, Pratul P. Srinivasan, Ben Mildenhall, Jonathan T. Barron, and Paul E. Debevec. Baking neural radiance fields for real-time view synthesis. In Proceedings of ICCV, pages 5855–5864. IEEE, 2021. 1

    2021

  16. [24]

    Hdr-nerf: High dynamic range neural radi- ance fields

    Xin Huang, Qi Zhang, Ying Feng, Hongdong Li, Xuan Wang, and Qing Wang. Hdr-nerf: High dynamic range neural radi- ance fields. In Proceedings of CVPR, pages 18377–18387. IEEE, 2022. 1

    2022

  17. [25]

    Slabaugh, Chunjing Xu, Ya-Li Li, Shengjin Wang, and Qi Tian

    Takashi Isobe, Songjiang Li, Xu Jia, Shanxin Yuan, Gre- gory G. Slabaugh, Chunjing Xu, Ya-Li Li, Shengjin Wang, and Qi Tian. Video super-resolution with temporal group attention. In Proceedings of CVPR, pages 8005–8014. Com- puter Vision Foundation / IEEE, 2020. 2

    2020

  18. [26]

    Multi-scale progressive fusion network for single image deraining

    Kui Jiang, Zhongyuan Wang, Peng Yi, Chen Chen, Baojin Huang, Yimin Luo, Jiayi Ma, and Junjun Jiang. Multi-scale progressive fusion network for single image deraining. In Proceedings of CVPR, pages 8343–8352. Computer Vision Foundation / IEEE, 2020. 2

    2020

  19. [27]

    Barron, Zhangyang Wang, and Tianfan Xue

    Yifan Jiang, Peter Hedman, Ben Mildenhall, Dejia Xu, Jonathan T. Barron, Zhangyang Wang, and Tianfan Xue. Alignerf: High-fidelity neural radiance fields via alignment- aware training. In Proceedings of CVPR, pages 46–55. IEEE,

  20. [28]

    Geonerf: Generalizing nerf with geometry priors

    Mohammad Mahdi Johari, Yann Lepoittevin, and Franc ¸ois Fleuret. Geonerf: Generalizing nerf with geometry priors. In Proceedings of CVPR, pages 18344–18347. IEEE, 2022. 1, 2, 5, 6, 7, 8, 13

    2022

  21. [29]

    Hdr-plenoxels: Self-calibrating high dynamic range radiance fields

    Kim Jun-Seong, Kim Yu-Ji, Moon Ye-Bin, and Tae-Hyun Oh. Hdr-plenoxels: Self-calibrating high dynamic range radiance fields. In Proceedings of ECCV, pages 384–401. Springer,

  22. [30]

    3d common corruptions and data augmentation

    Oguzhan Fatih Kar, Teresa Yeo, Andrei Atanov, and Amir Zamir. 3d common corruptions and data augmentation. In Proceedings of CVPR, pages 18941–18952. IEEE, 2022. 2, 4

    2022

  23. [31]

    Iterative residual cnns for burst photography applications

    Filippos Kokkinos and Stamatios Lefkimmiatis. Iterative residual cnns for burst photography applications. In Proceed- ings of CVPR, pages 5929–5938. Computer Vision Founda- tion / IEEE, 2019. 2

    2019

  24. [32]

    Deblurgan: Blind motion deblurring using conditional adversarial networks

    Orest Kupyn, V olodymyr Budzan, Mykola Mykhailych, Dmytro Mishkin, and Jiri Matas. Deblurgan: Blind motion deblurring using conditional adversarial networks. In Pro- ceedings of CVPR, pages 8183–8192. Computer Vision Foun- dation / IEEE Computer Society, 2018. 2

    2018

  25. [33]

    Deblurgan-v2: Deblurring (orders-of-magnitude) faster and better

    Orest Kupyn, Tetiana Martyniuk, Junru Wu, and Zhangyang Wang. Deblurgan-v2: Deblurring (orders-of-magnitude) faster and better. In Proceedings of ICCV, pages 8877–8886. IEEE,

  26. [34]

    Dp-nerf: Deblurred neural radiance field with physical scene priors

    Dogyoon Lee, Minhyeok Lee, Chajin Shin, and Sangyoun Lee. Dp-nerf: Deblurred neural radiance field with physical scene priors. In Proceedings of CVPR, pages 12386–12396. IEEE, 2023. 1, 2

    2023

  27. [35]

    Seathru- nerf: Neural radiance fields in scattering media

    Deborah Levy, Amit Peleg, Naama Pearl, Dan Rosenbaum, Derya Akkaynak, Simon Korman, and Tali Treibitz. Seathru- nerf: Neural radiance fields in scattering media. In Proceed- ings of CVPR, pages 56–65. IEEE, 2023. 1

    2023

  28. [36]

    Deep random projector: Accelerated deep image prior

    Taihui Li, Hengkang Wang, Zhong Zhuang, and Ju Sun. Deep random projector: Accelerated deep image prior. In Proceed- ings of CVPR, pages 18176–18185. IEEE, 2023. 4

    2023

  29. [37]

    On ef- ficient transformer-based image pre-training for low-level vision

    Wenbo Li, Xin Lu, Shengju Qian, and Jiangbo Lu. On ef- ficient transformer-based image pre-training for low-level vision. In Proceedings of IJCAI, pages 1089–1097. ijcai.org,

  30. [38]

    Swinir: Image restoration 10 using swin transformer

    Jingyun Liang, Jiezhang Cao, Guolei Sun, Kai Zhang, Luc Van Gool, and Radu Timofte. Swinir: Image restoration 10 using swin transformer. In Proceedings of ICCV Workshops, pages 1833–1844. IEEE, 2021. 2

    2021

  31. [39]

    BARF: bundle-adjusting neural radiance fields

    Chen-Hsuan Lin, Wei-Chiu Ma, Antonio Torralba, and Simon Lucey. BARF: bundle-adjusting neural radiance fields. In Proceedings of ICCV, pages 5721–5731. IEEE, 2021. 13

    2021

  32. [40]

    Neural rays for occlusion-aware image-based rendering

    Yuan Liu, Sida Peng, Lingjie Liu, Qianqian Wang, Peng Wang, Christian Theobalt, Xiaowei Zhou, and Wenping Wang. Neural rays for occlusion-aware image-based rendering. In Proceedings of CVPR, pages 7814–7823. IEEE, 2022. 2

    2022

  33. [41]

    Degae: A new pretraining paradigm for low- level vision

    Yihao Liu, Jingwen He, Jinjin Gu, Xiangtao Kong, Yu Qiao, and Chao Dong. Degae: A new pretraining paradigm for low- level vision. In Proceedings of CVPR, pages 23292–23303. IEEE, 2023. 2

    2023

  34. [42]

    Li Ma, Xiaoyu Li, Jing Liao, Qi Zhang, Xuan Wang, Jue Wang, and Pedro V . Sander. Deblur-nerf: Neural radiance fields from blurry images. In Proceedings of CVPR, pages 12851–12860. IEEE, 2022. 1, 2, 5, 6

    2022

  35. [43]

    Ricardo Martin-Brualla, Noha Radwan, Mehdi S. M. Saj- jadi, Jonathan T. Barron, Alexey Dosovitskiy, and Daniel Duckworth. Nerf in the wild: Neural radiance fields for un- constrained photo collections. In Proceedings of CVPR, pages 7210–7219. Computer Vision Foundation / IEEE, 2021. 2

    2021

  36. [44]

    Barron, Jiawen Chen, Dillon Sharlet, Ren Ng, and Robert Carroll

    Ben Mildenhall, Jonathan T. Barron, Jiawen Chen, Dillon Sharlet, Ren Ng, and Robert Carroll. Burst denoising with kernel prediction networks. In Proceedings of CVPR, pages 2502–2510. Computer Vision Foundation / IEEE Computer Society, 2018. 2, 4, 13

  37. [45]

    Srinivasan, Rodrigo Ortiz Cayon, Nima Khademi Kalantari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar

    Ben Mildenhall, Pratul P. Srinivasan, Rodrigo Ortiz Cayon, Nima Khademi Kalantari, Ravi Ramamoorthi, Ren Ng, and Abhishek Kar. Local light field fusion: practical view syn- thesis with prescriptive sampling guidelines. ACM Trans. Graph., 38(4):29:1–29:14, 2019. 6, 7

    2019

  38. [46]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view synthe- sis. In Proceedings of ECCV, pages 405–421. Springer, 2020. 1

    2020

  39. [47]

    Srinivasan, and Jonathan T

    Ben Mildenhall, Peter Hedman, Ricardo Martin-Brualla, Pratul P. Srinivasan, and Jonathan T. Barron. Nerf in the dark: High dynamic range view synthesis from noisy raw images. In Proceedings of CVPR, pages 16169–16178. IEEE,

  40. [48]

    Instant neural graphics primitives with a multiresolu- tion hash encoding

    Thomas M¨uller, Alex Evans, Christoph Schied, and Alexander Keller. Instant neural graphics primitives with a multiresolu- tion hash encoding. ACM Trans. Graph., 41(4):102:1–102:15,

  41. [49]

    NAN: noise- aware nerfs for burst-denoising

    Naama Pearl, Tali Treibitz, and Simon Korman. NAN: noise- aware nerfs for burst-denoising. In Proceedings of CVPR, pages 12662–12671. IEEE, 2022. 2, 3, 4, 5, 6, 7, 8, 13, 18

    2022

  42. [50]

    Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps

    Christian Reiser, Songyou Peng, Yiyi Liao, and Andreas Geiger. Kilonerf: Speeding up neural radiance fields with thousands of tiny mlps. InProceedings of ICCV, pages 14315– 14325. IEEE, 2021. 1

    2021

  43. [51]

    Generalizable patch-based neural render- ing

    Mohammed Suhail, Carlos Esteves, Leonid Sigal, and Ameesh Makadia. Generalizable patch-based neural render- ing. In Proceedings of ECCV, pages 156–174. Springer, 2022. 1, 2

    2022

  44. [52]

    Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction. In Proceedings of CVPR, pages 5449–5459. IEEE, 2022. 1

    2022

  45. [53]

    Is attention all that nerf needs? In Proceedings of ICLR, 2023

    Mukund Varma T, Peihao Wang, Xuxi Chen, Tianlong Chen, Subhashini Venugopalan, and Zhangyang Wang. Is attention all that nerf needs? In Proceedings of ICLR, 2023. 1, 2, 6, 7, 8, 13, 17

    2023

  46. [54]

    Effi- cient view synthesis and 3d-based multi-frame denoising with multiplane feature representations

    Thomas Tanay, Ales Leonardis, and Matteo Maggioni. Effi- cient view synthesis and 3d-based multi-frame denoising with multiplane feature representations. In Proceedings of CVPR, pages 20898–20907. IEEE, 2023. 2

    2023

  47. [55]

    Lempitsky

    Dmitry Ulyanov, Andrea Vedaldi, and Victor S. Lempitsky. Deep image prior. CoRR, abs/1711.10925, 2017. 4

    2017 arXiv

  48. [56]

    Yuille, and Cihang Xie

    Chen Wang, Angtian Wang, Junbo Li, Alan L. Yuille, and Cihang Xie. Benchmarking robustness in neural radiance fields. CoRR, abs/2301.04075, 2023. 2

    2023 arXiv

  49. [57]

    Patchmatchnet: Learned multi- view patchmatch stereo

    Fangjinhua Wang, Silvano Galliani, Christoph V ogel, Pablo Speciale, and Marc Pollefeys. Patchmatchnet: Learned multi- view patchmatch stereo. In Proceedings of CVPR , pages 14194–14203. Computer Vision Foundation / IEEE, 2021. 4

    2021

  50. [58]

    Bad- nerf: Bundle adjusted deblur neural radiance fields

    Peng Wang, Lingzhe Zhao, Ruijie Ma, and Peidong Liu. Bad- nerf: Bundle adjusted deblur neural radiance fields. In Pro- ceedings of CVPR, pages 4170–4179. IEEE, 2023. 1, 2

    2023

  51. [59]

    Srini- vasan, Howard Zhou, Jonathan T

    Qianqian Wang, Zhicheng Wang, Kyle Genova, Pratul P. Srini- vasan, Howard Zhou, Jonathan T. Barron, Ricardo Martin- Brualla, Noah Snavely, and Thomas A. Funkhouser. Ibrnet: Learning multi-view image-based rendering. In Proceedings of CVPR, pages 4690–4699. Computer Vision Foun...

    2021

  52. [60]

    Nerf-: Neural radiance fields without known camera parameters

    Zirui Wang, Shangzhe Wu, Weidi Xie, Min Chen, and Vic- tor Adrian Prisacariu. Nerf-: Neural radiance fields without known camera parameters. CoRR, abs/2102.07064, 2021. 2

    2021 arXiv

  53. [61]

    Nex: Real-time view synthesis with neural basis expansion

    Suttisak Wizadwongsa, Pakkapon Phongthawee, Jiraphon Yenphraphai, and Supasorn Suwajanakorn. Nex: Real-time view synthesis with neural basis expansion. In Proceedings of CVPR, pages 8534–8543. Computer Vision Foundation / IEEE, 2021. 1

    2021

  54. [62]

    Rafe: Generative radiance fields restoration, 2024

    Zhongkai Wu, Ziyu Wan, Jing Zhang, Jing Liao, and Dong Xu. Rafe: Generative radiance fields restoration, 2024. 2

    2024

  55. [63]

    Basis prediction networks for effective burst denoising with large kernels

    Zhihao Xia, Federico Perazzi, Micha ¨el Gharbi, Kalyan Sunkavalli, and Ayan Chakrabarti. Basis prediction networks for effective burst denoising with large kernels. In Proceed- ings of CVPR, pages 11841–11850. Computer Vision Foun- dation / IEEE, 2020. 2, 4, 13

    2020

  56. [64]

    Cf-nerf: Camera parame- ter free neural radiance fields with incremental learning

    Qingsong Yan, Qiang Wang, Kaiyong Zhao, Jie Chen, Bo Li, Xiaowen Chu, and Fei Deng. Cf-nerf: Camera parame- ter free neural radiance fields with incremental learning. In Proceedings of AAAI, pages 6440–6448. AAAI Press, 2024. 13

    2024

  57. [65]

    Mvsnet: Depth inference for unstructured multi-view stereo

    Yao Yao, Zixin Luo, Shiwei Li, Tian Fang, and Long Quan. Mvsnet: Depth inference for unstructured multi-view stereo. In Proceedings of ECCV, pages 785–801. Springer, 2018. 4

    2018

  58. [66]

    Underwater light field retention: Neural rendering for underwater imaging

    Tian Ye, Sixiang Chen, Yun Liu, Yi Ye, Erkang Chen, and Yuche Li. Underwater light field retention: Neural rendering for underwater imaging. In Proceedings of CVPR , pages 487–496. IEEE, 2022. 1 11

    2022

  59. [67]

    pixelnerf: Neural radiance fields from one or few images

    Alex Yu, Vickie Ye, Matthew Tancik, and Angjoo Kanazawa. pixelnerf: Neural radiance fields from one or few images. In Proceedings of CVPR, pages 4578–4587. Computer Vision Foundation / IEEE, 2021. 1, 2

    2021

  60. [68]

    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 CVPR, pages 5718–5729. IEEE,

  61. [69]

    Blur interpolation transformer for real-world motion from blur

    Zhihang Zhong, Mingdeng Cao, Xiang Ji, Yinqiang Zheng, and Imari Sato. Blur interpolation transformer for real-world motion from blur. In Proceedings of CVPR, pages 5713–5723. IEEE, 2023. 2

    2023

  62. [70]

    Occlusion-free scene recovery via neural radiance fields

    Chengxuan Zhu, Renjie Wan, Yunkai Tang, and Boxin Shi. Occlusion-free scene recovery via neural radiance fields. In Proceedings of CVPR, pages 20722–20731. IEEE, 2023. 2 12 A. Appendix Section Fig. 1 includes additional examples of our rendered images. For a better overview of...

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.