Pith. sign in

REVIEW 3 major objections 5 minor 27 references

IDArb: Intrinsic Decomposition for Arbitrary Number of Input Views and Illuminations

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read The paper's model, IDArb, decomposes any number of images taken under different lights into consistent albedo, normal, metallic, and roughness maps in a single feed-forward pass.

desk verdict Solid systems paper with a promising dataset and training recipe, but the headline SOTA numbers rest on an unstated evaluation split from the training distribution. read the letter →

arxiv 2412.12083 v3 pith:7VFGAZCI submitted 2024-12-16 cs.CV

classification cs.CV
keywords intrinsicdecompositioninverserenderingdiffusionmodelmulti-viewconsistencymaterialestimationrelightingphotometricstereoBRDF
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

IDArb claims that a single feed-forward diffusion model can decompose any number of input images of an object, captured under different and unknown lighting, into mutually consistent intrinsic maps: albedo, surface normal, metallic, and roughness. The load-bearing idea is that multi-view consistency and lighting disentanglement can be learned from data rather than enforced by optimization, so the same model handles one image or many images without knowing camera poses. If the claim holds, inverse rendering becomes a fast one-shot operation for relighting, material editing, photometric stereo, and 3D reconstruction, replacing hours of per-scene optimization and avoiding the inconsistent results of running single-image models repeatedly. The paper reports that the model outperforms existing single-image intrinsic decomposition and normal-estimation methods on every reported metric, and that adding views and lighting variations improves accuracy, especially for metallic and roughness.

What carries the argument

The mechanism that carries the argument is a cross-view, cross-component attention block added to the U-Net of a pretrained latent diffusion model. Within the attention operation, keys and values are gathered from all input views and from all three intrinsic triplets simultaneously, so each denoising step can exchange information about geometry, albedo, metal, and roughness across viewpoints. This is supported by an illumination-augmented, view-adaptive training schedule: images are sampled from different viewpoints and lighting conditions, and the model alternates between single-image and multi-image batches so that it retains material priors while learning multi-view fusion. A zero-terminal-SNR noise schedule shifted toward high noise levels is used because the structure of metallic and roughness maps differs from natural images.

What would settle it

Run the trained model on a collection of real objects with measured per-pixel metallic and roughness maps that vary across the surface, such as corroded bronze, and compare error against synthetic objects with global values; a large error gap confined to spatially varying materials would show that the synthetic-data assumption fails.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that intrinsic decomposition can be posed as a conditional diffusion-generation problem over an arbitrary number of views: sample $X_{1:N} \sim p(X_{1:N} \mid I_{1:N})$ from an arbitrary number of RGB images captured under unconstrained illumination. The intrinsic components are divided into three triplets (albedo, normal, and metallic with roughness), each denoised as a separate latent channel grouped with the others, and the diffusion U-Net is modified so attention runs both across views and across components. This lets the model jointly reason about shape, material, and lighting, which the authors say removes the ambiguity that makes a black pixel either dark albedo or shadow. Trained on a newly built 5.7-million-image synthetic dataset with multiple HDR environment lights and point lights per viewpoint, and with a training schedule that alternates one-view and multi-view inputs, the model claims to decompose from 1 to 12 viewpoints with consistent outputs and to improve photometric stereo and optimization-based inverse rendering as a prior.

Load-bearing premise

The whole approach rests on synthetic objects with mostly uniform metallic and roughness values being representative enough of real-world materials, so a model trained on them can estimate spatially varying materials on real objects.

Editorial extensions

If this is right

  • With a trained IDArb, a user can feed one image or many images under unknown lights and receive albedo, normal, metallic, and roughness maps that agree across views, enabling direct material editing and relighting.
  • Adding viewpoints and lighting conditions improves decomposition quality, especially for metallic and roughness, with gains diminishing beyond roughly eight viewpoints.
  • The model can act as a prior for optimization-based inverse rendering by supplying pseudo-material labels, reducing the problem of lighting baked into albedo and improving relighting quality.
  • The same unmodified model handles photometric stereo from one-light-at-a-time images and outperforms a dedicated photometric-stereo method in the paper's experiments.
  • Because training uses only object-centric synthetic data, real-world generalization is shown qualitatively, with known failures on outdoor scenes, text, and fine material detail.

Reading between the lines

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

  • Beyond the paper, the same architecture could be retrained on datasets with per-pixel spatially varying metal and roughness; the paper's own limitation suggests this is the next test for real-world generality.
  • Beyond the paper, the $O(N^2)$ cost of concatenating all views in cross-view attention implies a practical ceiling around a dozen or so views; sparse or hierarchical attention would be needed for dense capture.
  • Beyond the paper, multi-illumination training suggests a controllable experiment: train two models, one with varied lights and one with a single light, and compare metallic and roughness accuracy under cast shadows, which would isolate the contribution of photometric cues.
  • Beyond the paper, because the model is view-count agnostic, video frames of a rotating object under moving lights could be decomposed frame-by-frame with cross-view attention providing temporal consistency; the paper does not test this.
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

3 major / 5 minor

Summary. The paper proposes IDArb, a diffusion-based model for intrinsic decomposition (albedo, normal, metallic, roughness) that takes an arbitrary number of input RGB images captured under varying, unconstrained illumination and outputs multi-view consistent intrinsic maps. The method builds on Stable Diffusion by adding cross-view and cross-component attention modules, an illumination-augmented training strategy, and a view-adapted training scheme that alternates between single-view and 3-view inputs. The authors also introduce ARB-Objaverse, a synthetic dataset of 5.7M multi-view renderings with lighting variation, and evaluate IDArb on synthetic benchmarks, real-world images, MIT-Intrinsic, Stanford-ORB, photometric stereo, and optimization-based inverse rendering. The main quantitative claim is that IDArb outperforms state-of-the-art single-image and multi-view methods on all intrinsic decomposition metrics.

Significance. If the quantitative claims hold, IDArb would be a notable advance: it is a feed-forward model that handles a variable number of views, integrates cross-view and cross-component reasoning, and enables downstream relighting, photometric stereo, and 3D reconstruction without per-scene optimization. The ARB-Objaverse dataset is a large-scale resource that could benefit the inverse rendering community. The authors are also transparent about limitations, such as oversimplification of spatially varying materials and O(N^2) attention complexity. However, the significance is currently tempered by evaluation gaps: the primary synthetic benchmark appears to draw objects from the same distribution as training, multi-view consistency is claimed but not quantified, and one real-world benchmark shows mixed results. These issues do not invalidate the approach but must be addressed before the central performance claims can be accepted.

major comments (3)
  1. [Section 4.1 (Evaluation Dataset) and Table 1] The main quantitative result, Table 1, is potentially circular. The evaluation set is described as '441 objects from Arb-Objaverse and G-Objaverse', but these are also the sources of the training data (Section 3.2). The paper does not state whether the 441 objects are disjoint from the training objects, how the split was made, or whether any object-level separation was enforced. Since the model is trained on 68k+ Objaverse objects, random sampling without a held-out split could easily include training objects, which would make the SOTA gains in Table 1 self-referential. Please specify the exact object-level train/eval split and, ideally, report results on an independent synthetic benchmark or a manually verified subset with no overlap.
  2. [Section 3.3 and Section 4.2] The paper repeatedly claims 'multi-view consistent estimation' and 'multi-view consistency' as a central contribution, but no quantitative consistency metric is ever reported. The ablations in Figure 7 and Tables 4-7 measure per-view accuracy as a function of the number of views, not whether the outputs agree across views (e.g., pairwise albedo/normal differences on overlapping surfaces, reprojection error, or 3D consistency). Without such a metric, the multi-view consistency claim is supported only by qualitative figures. Please add a numerical consistency evaluation, such as average pairwise error between predictions for the same surface point seen from different views.
  3. [Appendix D, Table 8 and Section 4.2] The real-world benchmark results are mixed and appear to contradict the unqualified claim of superiority in Section 4.2. On MIT-Intrinsic, IntrinsicAnything achieves higher SSIM (0.896) than IDArb (0.876), while IDArb achieves better PSNR and LPIPS. The paper's statement that IDArb 'outperforms' IntrinsicAnything on real data is therefore not universally supported by the reported metrics. Please report all three metrics in the main text, discuss the SSIM gap, and either temper the claim or provide analysis explaining why PSNR/LPIPS improvements should be weighted more heavily.
minor comments (5)
  1. [Section 4.1] Specify how many of the 441 evaluation objects came from Arb-Objaverse versus G-Objaverse, and confirm that the four viewpoints per object are distinct from any views used during training.
  2. [Section 4.3] The sentence 'We analysis the effects of the number of viewpoints and lighting conditions' contains a grammatical error; it should read 'We analyze...'.
  3. [Figure 7 and Tables 4-7] The claim that increasing lighting conditions 'generally improves' performance is not monotonic in the tables; for example, in Table 6 metallic error at #V=2 increases from 0.061 (#L=1) to 0.068 (#L=2). Please either show statistical significance or qualify the claim as a trend rather than a uniform improvement.
  4. [Section 3.3 (Architecture)] The paper states that cross-view attention is 'proposed' here but also says the module is 'adopted' from Wonder3D in Section 1. Please clarify exactly which components are adapted from Wonder3D and which are novel to IDArb, to avoid ambiguity about the contribution.
  5. [Figure 9 caption] There is a typo in the caption: 'NVDiffecMC' should be 'NVDiffRecMC'.

Circularity Check

1 steps flagged · score 4.0 of 10

Synthetic benchmark is drawn from the same dataset pool used for training, so the headline SOTA claim is partly in-sample; independent real-world benchmarks limit the circularity burden.

  1. fitted input called prediction [Section 3.2 (ARB-Objaverse dataset) and Section 4.1 (Evaluation Dataset), Table 1]
    "For synthetic data, we sample 441 objects from Arb-Objaverse and G-Objaverse, selecting four viewpoints for each object. ... We select 68k 3D models from Objaverse ... For training, we further enhance the variability by combining this dataset with G-Objaverse and ABO."

    The model is trained on ARB-Objaverse plus G-Objaverse, and the synthetic evaluation set is sampled from exactly those two datasets. The paper never states that the 441 evaluated objects were held out from the 68k-object training pool, so the Table 1 comparison is, as presented, an in-sample fit: the 'predictions' on these objects are measured on the same distribution and object pool used to train the model. The headline claim that IDArb 'consistently achieves the best results among all albedo, normal, metallic and roughness metrics' therefore reduces, for the synthetic benchmark, to reporting numbers from a model trained on the evaluation pool against baselines that were not trained on that pool. This is the pattern of a fitted input being reported as a prediction.

full rationale

IDArb's core contribution is an empirical feed-forward diffusion model, so most of the derivation chain consists of architectural and training choices (cross-view attention, cross-component attention, illumination augmentation, view-adapted training, noise-schedule shifting) rather than first-principles claims that could be circular by definition. The one substantive circularity concern is the synthetic evaluation: the 441-object benchmark is sampled from the same two datasets used for training (ARB-Objaverse and G-Objaverse), and no held-out split is stated, making the Table 1 SOTA claim partly self-referential as presented. This is partially mitigated by independent real-world benchmarks and qualitative results, and the paper's own Limitations section concedes that synthetic global material values may oversimplify real objects. A further comparison-equity concern is that IID, IntrinsicAnything, and GeoWizard are evaluated with pretrained weights while RGB↔X was retrained on the authors' dataset, but this is a benchmarking fairness issue rather than circularity. Overall, the central claim still has independent content, so the score is 4 rather than higher.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper contributes an empirical system. It rests on a pretrained generative model, a self-built synthetic dataset, and several hand-chosen training decisions. The axioms above are the load-bearing domain assumptions; no new physical entities are introduced.

free parameters (4)
  • Number of input views during training (N) = 3 and 1, randomly sampled
    Chosen by hand in Section 3.3. The claim of supporting arbitrary N is an extrapolation, since no training example uses more than 3 views.
  • Training resolution = 256x256
    Fixed in Section 4.1; evaluation is at 512x512, so the model is used above its training resolution.
  • Noise scheduler shift and zero terminal SNR = No numeric value given in paper
    Adopted from Shi et al. 2023 and Lin et al. 2024; the ablation in Section 4.3 shows it is important for metallic and roughness, but the exact schedule is not specified.
  • Dataset construction quantities = 68k objects, 12 views, 7 illuminations
    Hand-chosen in Section 3.2; no sensitivity analysis is provided for these numbers.
assumptions (4)
  • domain assumption Disney BRDF with albedo, metallic, and roughness is a sufficient appearance model for the inputs the method will see.
    Section 3.1 and A.2 define the image formation model; all outputs are limited to these parameters.
  • domain assumption Models trained purely on synthetic Objaverse renderings will generalize to real-world photos.
    Generalization to real images is claimed in Section 4.2 and Appendix D despite synthetic-only training.
  • ad hoc to paper Appearance-based cross-view attention can find correspondences across arbitrary views without camera pose information.
    Section 3.3 concatenates view features and applies attention with no pose conditioning; the validity of this is not separately measured.
  • ad hoc to paper Synthetic objects with mostly global metallic and roughness values are representative of real materials.
    The paper's Limitations section states that most synthetic data use global metalness and roughness, which may oversimplify real objects.

how reviews work

0 comments
Cite this review

Pith. "Pith review of IDArb: Intrinsic Decomposition for Arbitrary Number of Input Views and Illuminations." pith.science (2026). https://pith.science/paper/7VFGAZCI

@misc{pith2026241212083,
  author       = {Pith},
  title        = {Pith review of: IDArb: Intrinsic Decomposition for Arbitrary Number of Input Views and Illuminations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7VFGAZCI}},
  note         = {Machine review of arXiv:2412.12083}
}
read the original abstract

Capturing geometric and material information from images remains a fundamental challenge in computer vision and graphics. Traditional optimization-based methods often require hours of computational time to reconstruct geometry, material properties, and environmental lighting from dense multi-view inputs, while still struggling with inherent ambiguities between lighting and material. On the other hand, learning-based approaches leverage rich material priors from existing 3D object datasets but face challenges with maintaining multi-view consistency. In this paper, we introduce IDArb, a diffusion-based model designed to perform intrinsic decomposition on an arbitrary number of images under varying illuminations. Our method achieves accurate and multi-view consistent estimation on surface normals and material properties. This is made possible through a novel cross-view, cross-domain attention module and an illumination-augmented, view-adaptive training strategy. Additionally, we introduce ARB-Objaverse, a new dataset that provides large-scale multi-view intrinsic data and renderings under diverse lighting conditions, supporting robust training. Extensive experiments demonstrate that IDArb outperforms state-of-the-art methods both qualitatively and quantitatively. Moreover, our approach facilitates a range of downstream tasks, including single-image relighting, photometric stereo, and 3D reconstruction, highlighting its broad applications in realistic 3D content creation.

Figures

Figures reproduced from arXiv: 2412.12083 by the authors.

Figure 1
Figure 1. IDArb tackles intrinsic decomposition for an arbitrary number of views under un￾constrained illumination. Our approach (a) achieves multi-view consistency compared to learning￾based methods and (b) effectively disentangles intrinsic components from lighting effects compared to optimization-based methods. Our method enhances a wide range of applications such as image editing, photometric stereo, and 3D reconstruction… view at source ↗
Figure 2
Figure 2. Top: Overview of IDArb. Bottom: Illustration of the attention block within the UNet. Our training batch consists of N input images, sampled from Nv viewpoints and Ni illuminations. The latent vector for each image is concatenated with Gaussian noise for denoising. Intrinsic com￾ponents are divided into three triplets (D=3): Albedo, Normal and Metallic&Roughness. Specific text prompts are used to guide the model towa… view at source ↗
Figure 3
Figure 3. Overview of the Arb-Objaverse dataset. Our custom dataset features a diverse collection of objects rendered under various lighting conditions, accompanied by their intrinsic components. 3.3 ARCHITECTURE AND TRAINING Given an arbitrary number of views from single to multi-view images, IDArb generates multi-view consistent intrinsic maps under unconstrained illumination using a text-guided diffusion model. We base our… view at source ↗
Figures from the paper (17 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison on synthetic data. IDArb demonstrates superior intrinsic esti￾mation compared to all other methods [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison on real-world data. Input w/o cross-component attention w/ cross-component attention Ground truth Albedo Normal Metallic Roughness (a) Ablation on cross-component attention. Input w/o single view training w/o shift scheduler Ours Albedo Normal Me…
Figure 6
Figure 6. Figure 6: Ablative studies on (a) cross-component attention and (b) training strategy. leading to a loss of fidelity. In contrast, our model generates accurate and convincing decompositions with preserved details. Despite being trained on synthetic data, IDArb generalizes well t…
Figure 7
Figure 7. Figure 7: Effects of number of viewpoints and lighting conditions. We find increasing the number of viewpoints and the lighting conditions generally improves decomposition performance [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Relighting and material editing results. From in-the-wild captures (a), our model allows for relighting under novel illumination (b) and material property modifications (c). Photometric stereo. Photometric stereo is a long-standing challenge in computer vision, aiming …
Figure 9
Figure 9. Figure 9: Optimization-based inverse rendering results. Our method guides NVDiffecMC gen￾erate more plausible material results [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 13
Figure 13. Figure 13: 15 [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]
Figure 10
Figure 10. Figure 10: Photometric stereo results using 4 OLAT images in OpenIllumination and NeRFactor [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]
Figure 11
Figure 11. Figure 11: Qualitative comparison on MIT-Intrinsic (Grosse et al., 2009) with Intrinsi￾cAnything (Chen et al., 2024). Input image and ground truth have been contrast-adjusted for better visibility. features of a telephone. This issue arises from the synthetic training data, whic…
Figure 12
Figure 12. Figure 12: More results on real-world data. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_12.png]
Figure 13
Figure 13. Figure 13: More results on real-world data. We also provide the reconstructed and relighting images. [PITH_FULL_IMAGE:figures/full_fig_p019_13.png]
Figure 14
Figure 14. Figure 14: More results on multi-view data. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_14.png]
Figure 15
Figure 15. Figure 15: Multiview images with extreme lighting variation. For each scene in NeRD dataset (Boss et al., 2021a), we input 4 views. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_15.png]
Figure 16
Figure 16. Figure 16: Failure cases. 22 [PITH_FULL_IMAGE:figures/full_fig_p022_16.png]
Figure 17
Figure 17. Figure 17: Results on Mip-NeRF 360 (Barron et al., 2022) (Part 1, outdoor). We input 4 views for each scene. 23 [PITH_FULL_IMAGE:figures/full_fig_p023_17.png]
Figure 18
Figure 18. Figure 18: Results on Mip-NeRF 360 (Barron et al., 2022) (Part 2, indoor). We input 4 views for each scene. 24 [PITH_FULL_IMAGE:figures/full_fig_p024_18.png]
Figure 19
Figure 19. Figure 19: Results on indoor and outdoor scenes. Input images are collected from the Internet. 25 [PITH_FULL_IMAGE:figures/full_fig_p025_19.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 16 canonical work pages

  1. [2]

    Deep 3D Capture: Geometry and Reflectance from Sparse Multi-View Images

    URLhttps://arxiv.org/ abs/2003.12642. Mark Boss, Varun Jampani, Kihwan Kim, Hendrik P.A. Lensch, and Jan Kautz. Two-shot spatially- varying brdf and shape estimation. In IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR),

  2. [3]

    Barron, Ce Liu, and Hendrik P

    Mark Boss, Raphael Braun, Varun Jampani, Jonathan T. Barron, Ce Liu, and Hendrik P. A. Lensch. Nerd: Neural reflectance decomposition from image collections. In ICCV, pp. 12664–12674. IEEE, 2021a. Mark Boss, Varun Jampani, Raphael Braun, Ce Liu, Jonathan T. Barron, and Hendrik P. A. Lensch. Neural-pil: Neural pre-integrated lighting for reflectance decomp...

  3. [7]

    Relightable 3d gaussian: Real-time point cloud relighting with brdf decomposition and ray tracing

    Jian Gao, Chun Gu, Youtian Lin, Hao Zhu, Xun Cao, Li Zhang, and Yao Yao. Relightable 3d gaussian: Real-time point cloud relighting with brdf decomposition and ray tracing. arXiv:2311.16043,

  4. [8]

    Ground truth dataset and baseline evaluations for intrinsic image algorithms

    Roger Grosse, Micah K Johnson, Edward H Adelson, and William T Freeman. Ground truth dataset and baseline evaluations for intrinsic image algorithms. In 2009 IEEE 12th International Confer- ence on Computer Vision, pp. 2335–2342. IEEE,

  5. [10]

    Xin Kong, Shikun Liu, Xiaoyang Lyu, Marwan Taher, Xiaojuan Qi, and Andrew J Davison

    URL https://arxiv.org/abs/2312.12274. Xin Kong, Shikun Liu, Xiaoyang Lyu, Marwan Taher, Xiaojuan Qi, and Andrew J Davison. Esch- ernet: A generative model for scalable view synthesis. arXiv preprint arXiv:2402.03908,

  6. [11]

    Learning to reconstruct shape and spatially-varying reflectance from a single image

    Zhengqin Li, Zexiang Xu, Ravi Ramamoorthi, Kalyan Sunkavalli, and Manmohan Chandraker. Learning to reconstruct shape and spatially-varying reflectance from a single image. In SIG- GRAPH Asia 2018 Technical Papers, pp

  7. [13]

    Isabella Liu, Linghao Chen, Ziyang Fu, Liwen Wu, Haian Jin, Zhong Li, Chin Ming Ryan Wong, Yi Xu, Ravi Ramamoorthi, Zexiang Xu, and Hao Su

    URL https://arxiv.org/abs/2305.08891. Isabella Liu, Linghao Chen, Ziyang Fu, Liwen Wu, Haian Jin, Zhong Li, Chin Ming Ryan Wong, Yi Xu, Ravi Ramamoorthi, Zexiang Xu, and Hao Su. Openillumination: A multi-illumination dataset for inverse rendering evaluation on real objects, 2024a. Ruoshi Liu, Rundi Wu, Basile Van Hoorick, Pavel Tokmakov, Sergey Zakharov, ...

  8. [14]

    Syncdreamer: Generating multiview-consistent images from a single-view image, 2024b

    Yuan Liu, Cheng Lin, Zijiao Zeng, Xiaoxiao Long, Lingjie Liu, Taku Komura, and Wenping Wang. Syncdreamer: Generating multiview-consistent images from a single-view image, 2024b. URL https://arxiv.org/abs/2309.03453. 12 Xiaoxiao Long, Yuan-Chen Guo, Cheng Lin, Yuan Liu, Zhiyang Dou, Lingjie Liu, Yuexin Ma, Song-Hai Zhang, Marc Habermann, Christian Theobalt...

Show all 27 references
  1. [16]

    Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction

    Peng Wang, Lingjie Liu, Yuan Liu, Christian Theobalt, Taku Komura, and Wenping Wang. Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction. arXiv preprint arXiv:2106.10689,

  2. [17]

    Shangzhe Wu, Ameesh Makadia, Jiajun Wu, Noah Snavely, Richard Tucker, and Angjoo Kanazawa

    URL https://arxiv.org/abs/2201.02279. Shangzhe Wu, Ameesh Makadia, Jiajun Wu, Noah Snavely, Richard Tucker, and Angjoo Kanazawa. De-rendering the world’s revolutionary artefacts,

  3. [19]

    Chongjie Ye, Lingteng Qiu, Xiaodong Gu, Qi Zuo, Yushuang Wu, Zilong Dong, Liefeng Bo, Yuliang Xiu, and Xiaoguang Han

    URL https: //arxiv.org/abs/2208.12697. Chongjie Ye, Lingteng Qiu, Xiaodong Gu, Qi Zuo, Yushuang Wu, Zilong Dong, Liefeng Bo, Yuliang Xiu, and Xiaoguang Han. Stablenormal: Reducing diffusion variance for stable and sharp normal. ACM Transactions on Graphics (TOG),

  4. [20]

    Zheng Zeng, Valentin Deschaintre, Iliyan Georgiev, Yannick Hold-Geoffroy, Yiwei Hu, Fujun Luan, Ling-Qi Yan, and Miloˇs Haˇsan

    URL https://arxiv.org/abs/2303.13852. Zheng Zeng, Valentin Deschaintre, Iliyan Georgiev, Yannick Hold-Geoffroy, Yiwei Hu, Fujun Luan, Ling-Qi Yan, and Miloˇs Haˇsan. RGB ↔ X: Image decomposition and synthesis using material- and lighting-aware diffusion models. arXiv preprint ...

  5. [21]

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala

    URL https://arxiv.org/abs/2406.13897. Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models,

  6. [22]

    Srinivasan, Boyang Deng, Paul Debevec, William T

    Xiuming Zhang, Pratul P. Srinivasan, Boyang Deng, Paul Debevec, William T. Freeman, and Jonathan T. Barron. Nerfactor: neural factorization of shape and reflectance under an unknown il- lumination. ACM Transactions on Graphics, 40(6):1–18, December 2021b. ISSN 1557-7368. doi: ...

  7. [23]

    URL https: //arxiv.org/abs/2206.08423. A P RELIMINARY A.1 I MAGE DIFFUSION MODEL In Denoising Diffusion Probabilistic Models (DDPM) (Ho et al., 2020), a forward diffusion process is defined, gradually introducing small amounts of Gaussian noise to the sample at each timestep, ...

  8. [25]

    For MIT-Intrinsic, we compared our albedo estimation results with IntrinsicAnything (Chen et al., 2024), as shown in Tab

    and Stanford-ORB (Kuang et al., 2023). For MIT-Intrinsic, we compared our albedo estimation results with IntrinsicAnything (Chen et al., 2024), as shown in Tab. 8 and Fig

  9. [26]

    and IntrinsicNeRF (Ye et al., 2023), as shown in Tab

  10. [27]

    Input image and ground truth have been contrast-adjusted for better visibility

    with Intrinsi- cAnything (Chen et al., 2024). Input image and ground truth have been contrast-adjusted for better visibility. features of a telephone. This issue arises from the synthetic training data, which often contains simpler material variations, leading the model to ove...

  11. [1986]

    For a point x with surface normal n, the incident light intensity at this point is denoted as Li(ωi; x), where ωi represents the incident light direction

    to ensure physical correctness. For a point x with surface normal n, the incident light intensity at this point is denoted as Li(ωi; x), where ωi represents the incident light direction. The Bidirectional Reflectance Dis- tribution Function (BRDF) (Nicodemus, 1965), denoted as...

  12. [2009]

    Shape, Light, and Material Decomposition from Images using Monte Carlo Rendering and Denoising

    11 Jon Hasselgren, Nikolai Hofmann, and Jacob Munkberg. Shape, Light, and Material Decomposition from Images using Monte Carlo Rendering and Denoising. arXiv:2206.03380,

  13. [2016]

    Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang, Minghua Liu, Chao Xu, Xinyue Wei, Linghao Chen, Chong Zeng, and Hao Su

    URL https://arxiv.org/abs/1612.08510. Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang, Minghua Liu, Chao Xu, Xinyue Wei, Linghao Chen, Chong Zeng, and Hao Su. Zero123++: a single image to consistent multi-view diffusion base model,

  14. [2019]

    Shanchuan Lin, Bingchen Liu, Jiashi Li, and Xiao Yang

    URL https://arxiv.org/abs/1905.02722. Shanchuan Lin, Bingchen Liu, Jiashi Li, and Xiao Yang. Common diffusion noise schedules and sample steps are flawed,

  15. [2020]

    Jonathan T

    URL https://arxiv.org/abs/2010.03592. Jonathan T. Barron, Ben Mildenhall, Dor Verbin, Pratul P. Srinivasan, and Peter Hedman. Mip-nerf 360: Unbounded anti-aliased neural radiance fields. CVPR,

  16. [2021]

    Tong Wu, Jiaqi Wang, Xingang Pan, Xudong Xu, Christian Theobalt, Ziwei Liu, and Dahua Lin

    URL https://arxiv.org/abs/ 2104.03954. Tong Wu, Jiaqi Wang, Xingang Pan, Xudong Xu, Christian Theobalt, Ziwei Liu, and Dahua Lin. V oxurf: V oxel-based efficient and accurate neural surface reconstruction,

  17. [2022]

    Xiao Fu, Wei Yin, Mu Hu, Kaixuan Wang, Yuexin Ma, Ping Tan, Shaojie Shen, Dahua Lin, and Xiaoxiao Long

    URL https://arxiv.org/abs/2212.08051. Xiao Fu, Wei Yin, Mu Hu, Kaixuan Wang, Yuexin Ma, Ping Tan, Shaojie Shen, Dahua Lin, and Xiaoxiao Long. Geowizard: Unleashing the diffusion priors for 3d geometry estimation from a single image. In ECCV,

  18. [2023]

    doi: 10.1145/3630750

    ISSN 1557-7368. doi: 10.1145/3630750. URL http://dx.doi.org/10.1145/3630750. Xi Chen, Sida Peng, Dongchen Yang, Yuan Liu, Bowen Pan, Chengfei Lv, and Xiaowei Zhou. Intrinsicanything: Learning diffusion priors for inverse rendering under unknown illumination,

  19. [2024]

    URL https://arxiv.org/abs/2404.11593. Jasmine Collins, Shubham Goel, Kenan Deng, Achleshwar Luthra, Leon Xu, Erhan Gundogdu, Xi Zhang, Tomas F Yago Vicente, Thomas Dideriksen, Himanshu Arora, Matthieu Guillaumin, and Jitendra Malik. Abo: Dataset and benchmarks for real-world 3...

Pith tools

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