Pith. sign in

REVIEW 3 major objections 5 minor 85 references

NeRF-Texture: Synthesizing Neural Radiance Field Textures

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

Pith's one-line read Real-world textures with fine 3D geometry can be captured as NeRF textures and synthesized to any size, then mapped onto new shapes.

desk verdict Solid method paper that adds a genuinely missing capability—NeRF texture synthesis—with an honest failure analysis; the main weaknesses are evaluation rigor and under-specified reproducibility, not the core idea. read the letter →

arxiv 2412.10004 v1 pith:YVD5SLAJ submitted 2024-12-13 cs.CV cs.GR

classification cs.CVcs.GR
keywords neuralradiancefieldstexturesynthesismeso-structureview-dependentappearancedisentangledrepresentationlatentpatchmatchinghashgridencodingsphericalharmonics
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 proposes NeRF-Texture, a pipeline that captures, models, synthesizes, and re-renders real-world textures that have fine 3D geometry (meso-structure) and view-dependent appearance. Its central claim is that such textures can be disentangled from a coarse base shape and stored as latent features on that shape, then synthesized to arbitrary size by patch-matching the latent features rather than pixel colors. If correct, a user could shoot a short video of a material, synthesize arbitrarily large 3D textures from it, and map them onto new shapes while preserving the meso-structure and its lighting effects. The paper presents this as the first NeRF texture synthesis method, extending its earlier conference version to curved surfaces.

What carries the argument

The central object is the disentangled NeRF texture representation: 3D space is reparameterized through a differentiable projection onto a coarse base mesh, giving a footpoint and a signed distance, and hash grids store latent texture features indexed by the footpoint. The argument is carried by combining this representation with implicit patch matching: patches of latent features are sampled on the base shape, matched and quilted in feature space by the minimum-cost-cut procedure of image quilting, and a Student t-distribution clustering loss regularizes the feature distribution so that feature-space distance tracks content-space distance. A custom derivation rule for the projection, dxc/dx = I - nc^T nc and ds/dx = nc, lets gradients flow through the footpoint for normal estimation and camera optimization. For curved surfaces, an atlas UV map and a coarse-to-fine patch pyramid replace the fixed kd-tree acceleration used in planar synthesis.

What would settle it

For a scene with a clean base shape, such as a flat wall of stones, render the synthesized texture from a grazing angle and measure SIFID against a real photograph: the paper reports 0.82 at 80 degrees elevation, so a substantially larger value would indicate that the view-dependent meso-structure is not actually preserved.

Watch

Extended reading notes

Core claim

NeRF-Texture represents a textured scene as a smooth base mesh plus a NeRF texture: two hash-grid feature fields defined on the base mesh's surface, queried by the footpoint of each 3D sample, together with the sample's signed distance from the mesh. An MLP decoder turns these features into density, diffuse and specular Phong coefficients, glossiness, and the elevation and azimuth of a fine normal expressed in the local tangent frame; color is then computed with spherical-harmonic lighting. Synthesis is performed by extracting implicit patches of latent features on the base shape, then running the image-quilting approach of patch matching and minimum-cut stitching directly in latent space, with a clustering loss that makes latent distance correlate with reconstructed content similarity. The synthesized latent field can be mapped onto any new mesh, including curved surfaces via an atlas parameterization, and rendered in real time. The paper demonstrates applications on bark, durian, fabric, leaves, flowers, mirror balls, a metal bed, and curved targets such as a ring, a shark, and a tower.

Load-bearing premise

The load-bearing premise is that a real textured scene can be split cleanly into a smooth base shape plus latent features on that shape that carry all meso-structure and appearance, so the whole pipeline inherits the quality of the base shape extraction.

Editorial extensions

If this is right

  • A single short video of a material can produce a reusable NeRF texture that is applied to many different shapes without re-training the model.
  • Synthesized NeRF textures preserve meso-structure occlusion and view-dependent reflection at high viewing angles, where 2D image textures visibly break down.
  • The method renders at around 84 FPS, making the synthesized textures practical for real-time applications rather than offline baking.
  • The same latent patch-matching pipeline extends to arbitrary curved surfaces through atlas parameterization with a coarse-to-fine matching strategy.

Reading between the lines

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

  • If the latent metric after clustering truly tracks content similarity, the clustering regularization could transfer to other neural fields that need patch-based editing, such as relightable or animatable radiance fields.
  • A natural testable extension is to measure whether nearest-neighbor search in the clustered latent space returns patches whose rendered appearance matches human judgment, which would validate the metric-consistency argument directly.
  • The representation effectively turns a 3D texture into a reusable asset with its own lighting decomposition, so a library of captured NeRF textures could be built and shared much like 2D texture libraries.
  • Building on the paper's own suggestion, adding a generative completion model could extend the method to thin or semantically meaningful structures like railings and keycaps, where the current greedy patch matching is known to break continuity.
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 / 5 minor

Summary. The paper presents NeRF-Texture, a method to capture, model, synthesize, and re-render textures with meso-structure (geometric detail) and view-dependent appearance from multi-view images. The scene is disentangled into a coarse base mesh and a latent feature field defined on that mesh; query points are parameterized by signed distance and footpoint, and a small MLP decodes density, shading coefficients, and fine normals. Texture synthesis is performed by implicit patch matching of latent features over a planar domain or curved surfaces, with a clustering regularizer to align latent-space and content-space metrics. Experiments include qualitative results on real-world captures, quantitative view synthesis on DTU, comparison with 2D textures via SIFID, comparison with NeRF-Tex, ablations, and a challenge analysis.

Significance. If the central claims hold, this is a useful and timely contribution to neural texture synthesis: it is among the first to treat meso-structure textures as neural radiance fields that can be captured from real images, synthesized, and mapped to new shapes, while preserving view-dependent effects. The paper provides a fairly detailed description of the representation and synthesis pipeline, and it is commendable that the authors include a challenge analysis (Sec. 5) that candidly identifies failure modes for narrow structures and complex topology. However, the quantitative support for key claims is thinner than the breadth of the results suggests: the SIFID comparison is reported without error bars, the clustering ablation is qualitative only, and the base-shape extraction assumption is not validated quantitatively. With additional experiments and clarifications, the contribution would be solid.

major comments (3)
  1. [Sec. 3.1.1 and Sec. 5.1] The base shape extraction procedure (Instant-NGP + Co-ACD + Laplacian smoothing/remeshing) is a generic heuristic with no principled scale-separation criterion. The signed distance s(x) and tangent frame Tc(xc) are defined relative to this base; if the extracted base is inaccurate (as the paper's own examples of truss and Lego cockpit in Sec. 5.1 show), the learned representation is biased and the decoder cannot transfer to new shapes. The paper does not quantify under what geometric conditions the disentanglement assumption holds, nor does it provide a diagnostic for detecting base-shape failure before synthesis. The challenge analysis lists failure cases but does not analyze the sensitivity of the representation to base-shape error. Please add a quantitative sensitivity study (e.g., perturbing the base mesh and measuring synthesis quality) and a discussion of the geometric properties required for the assumption to be valid.
  2. [Sec. 4.5 and Eq. (4)] The clustering loss Lclu is presented as a key contribution for improving patch matching, yet the ablation in Sec. 4.5 is purely qualitative: Fig. 17 shows PCA visualizations and a rendered comparison, but no quantitative metric. The claim that the clustering constraint 'reduces artifacts' is not substantiated with numbers. Please report a quantitative measure of synthesis quality with and without Lclu (e.g., SIFID on the synthesized textures, patch matching error, or a perceptual metric) across multiple textures, including error bars. This is load-bearing because the paper motivates Lclu as essential for matching performance.
  3. [Table 2 and Sec. 4.3] The SIFID comparison against 2D textures is the main quantitative evidence for the advantage of the NeRF-based representation, but Table 2 reports only a single average per elevation angle with no standard deviation, number of textures/scenes, or statistical test. The protocol for cropping ground-truth regions and selecting 'closest viewing directions' is not fully specified, making the comparison hard to reproduce or interpret. Please provide per-scene/per-texture results with error bars, and describe the ground-truth selection procedure in enough detail that the evaluation is reproducible. As written, the table does not convincingly support the claim that the NeRF-based representation is 'more realistic than 2D textures.'
minor comments (5)
  1. [Eq. (1)] The formula for the coarse normal n~c(x) is ambiguous: the second term (x - v1)/(w||x - v1||^2) appears inside the summation in the typeset equation, which would add it K times. Please place the term outside the summation or clarify the intended expression, and define W unambiguously.
  2. [Sec. 4.4] The comparison with NeRF-Tex is described only qualitatively, and the training data for NeRF-Tex is generated using the authors' own representation. Please provide more implementation details (e.g., bounding box size, number of views, lighting) so that the fairness of the comparison can be assessed, and consider reporting a quantitative metric.
  3. [Sec. 5.1] Figure references are out of order: the text mentions 'the second row of Fig. 21' before introducing Fig. 20, and the Lego example in Fig. 20 is referenced after Fig. 21. Please reorder the figures or the text so that the discussion matches the figure numbering.
  4. [Sec. 3.1.4] The claim that latent features are fetched in 'O(1) time complexity' is only true for the hash lookup itself; the overall pipeline includes KNN search and ray casting, which have higher complexity. Please qualify the statement to avoid overclaiming.
  5. [General] There are several typos and formatting issues, including 'Standord Bunny' (Sec. 4.1), 'F r e c h e t' (Sec. 4.3), and the malformed reference [4] ('Bao and Yang, Z. Junyi...'). A careful proofread is needed.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: the method's outputs are validated against external baselines and ground-truth crops, and the only self-citations are provenance or related-work pointers, not premises in the derivation.

full rationale

The paper does not present a derivation that reduces to its inputs. Its central procedure is an optimization: multi-view images are used to train a disentangled representation (coarse base mesh, hash-grid latent features, SDF, Phong/SH shading) with reconstruction, distortion, normal, and clustering losses (Sec. 3.3). The synthesized outputs are produced by patch matching and quilting of learned latent features (Sec. 3.2), a standard exemplar-based synthesis pipeline; the output is not a fitted quantity renamed as a prediction. Quantitative evaluation compares against external methods (NeRF, Instant-NGP, NeRF-Tex) and against ground-truth image crops using SIFID (Sec. 4.3), so the quality claims are not self-justified. The clustering loss (Eq. 4) is a regularizer whose effect is ablated in Sec. 4.5; it is not a parameter tuned to the evaluation metric. The only self-references are [8], the authors' own SIGGRAPH 2023 version, mentioned as provenance in Sec. 1, and [23], a related stylization paper; neither is invoked to justify an equation or to preclude alternative designs. Sec. 5.1 honestly documents failure modes (limited coarse extents and complex topology) that follow from the base-shape assumption; these are limitations, not circular reasoning. The score of 2 reflects only the presence of minor non-load-bearing self-citation in the manuscript; no circular step was found.

Assumptions & free parameters 11 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a series of hand-chosen hyperparameters and modeling assumptions, none of which are derived from first principles. The most consequential are the base shape decomposition, the rendering model, and the clustering regularizer. No new physical entities are introduced. The paper provides no formal proofs; correctness is an empirical matter.

free parameters (11)
  • Loss weights lambda_1, lambda_2, lambda_3 = 1e-5, 1e-2, 1
    Hand-set in Sec. 3.3. The clustering weight lambda_1 directly controls the latent regularization that the paper says improves synthesis, so the result depends on these values.
  • Patch resolution = 128x128
    Selected through ablation in Sec. 4.6. This is the granularity of all patch extraction and matching, so it is load-bearing for synthesis quality.
  • kNN neighbor count K = 8
    Used in Eq. (1) to compute coarse normals; fixed without sensitivity analysis.
  • Normal interpolation weight w = 0.01
    Constant in Eq. (1) balancing KNN normals and the point-to-mesh direction; no ablation.
  • Clustering degrees of freedom kappa = not reported
    Parameter of Student's t-distribution in Eq. (4); its value is not given, so the clustering behavior cannot be reproduced exactly.
  • Number of cluster centers = not reported
    Required to define the clustering loss; not specified in the paper.
  • Patch overlap width = not reported
    The quilting algorithm's overlap size is a key synthesis parameter, but it is not stated.
  • UV atlas resolution = 2048x2048
    Used for curved surface synthesis (Sec. 3.2.3) and affects detail retention and storage.
  • Number of patch samples on curved surfaces = 8000
    Stated in Sec. 3.2.3 as a sampling budget; this is a hand-tuned quantity to ensure enough candidates.
  • Hash grid parameters (resolution, levels, feature dimension) = not reported
    These determine the latent feature capacity and are not specified, hurting reproducibility.
  • Number of training views = 150 to 300
    Capture protocol in Sec. 4.7; with only 4-16 views the results degrade, so the pipeline requires this many views.
assumptions (5)
  • domain assumption A real scene with meso-structure can be decomposed into a smooth base shape and latent texture features attached to that base shape.
    This is the central modeling assumption of Sec. 3.1. The method's failures on complex topology (Sec. 5.1) show this does not always hold.
  • domain assumption The rendering equation with Phong shading and spherical-harmonic lighting is sufficient to represent real-world appearance.
    Sec. 3.1.5 assumes this decomposition. Effects outside the model, such as complex interreflections, are not captured, as the paper notes when adding Ref-NeRF for specular materials.
  • ad hoc to paper The custom differentiable projection rule in Eq. (2) correctly propagates gradients for training.
    The rule is a design choice, not a standard embedding. The optimization relies on it.
  • ad hoc to paper The clustering loss (Eq. 4) aligns latent-space distances with content-space distances, improving patch matching.
    Sec. 3.2.4 motivates this heuristically; only qualitative evidence is provided for its benefit.
  • domain assumption The base shape extracted via Instant-NGP and Co-ACD is a valid parameterization for any input texture.
    Sec. 3.1.1 relies on this. Narrow or topologically complex objects violate it, as shown in Sec. 5.1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of NeRF-Texture: Synthesizing Neural Radiance Field Textures." pith.science (2026). https://pith.science/paper/YVD5SLAJ

@misc{pith2026241210004,
  author       = {Pith},
  title        = {Pith review of: NeRF-Texture: Synthesizing Neural Radiance Field Textures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YVD5SLAJ}},
  note         = {Machine review of arXiv:2412.10004}
}
read the original abstract

Texture synthesis is a fundamental problem in computer graphics that would benefit various applications. Existing methods are effective in handling 2D image textures. In contrast, many real-world textures contain meso-structure in the 3D geometry space, such as grass, leaves, and fabrics, which cannot be effectively modeled using only 2D image textures. We propose a novel texture synthesis method with Neural Radiance Fields (NeRF) to capture and synthesize textures from given multi-view images. In the proposed NeRF texture representation, a scene with fine geometric details is disentangled into the meso-structure textures and the underlying base shape. This allows textures with meso-structure to be effectively learned as latent features situated on the base shape, which are fed into a NeRF decoder trained simultaneously to represent the rich view-dependent appearance. Using this implicit representation, we can synthesize NeRF-based textures through patch matching of latent features. However, inconsistencies between the metrics of the reconstructed content space and the latent feature space may compromise the synthesis quality. To enhance matching performance, we further regularize the distribution of latent features by incorporating a clustering constraint. In addition to generating NeRF textures over a planar domain, our method can also synthesize NeRF textures over curved surfaces, which are practically useful. Experimental results and evaluations demonstrate the effectiveness of our approach.

Figures

Figures reproduced from arXiv: 2412.10004 by the authors.

Figure 1
Figure 1. Given a set of multi-view images of the target texture with meso-structure, our model synthesizes Neural Radiance Field (NeRF) textures, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of our method. Given a set of multi-view images, we first estimate its base shape. Based on it, we model the scene with a disentangled representation of the base shape and NeRF texture with meso-structure. The query point x is projected onto the base shape as footpoint xc. Latent features f(x), fˆ(x) representing textures are fetched by feeding xc to hash grids. Along with matrices of local tangent space Tc… view at source ↗
Figure 3
Figure 3. Base Shape Extraction. We show the intermediate outputs during the base shape extraction, including NGP [6], Co-ACD [7], and re-meshing [51], [52]. appearance. To disentangle these attributes and the base shape, we utilize the coarse mesh mentioned above to re￾parameterize 3D Euclidean space and learn the attributes into the latent features defined on the coarse mesh. In our approach, the coordinates of query point … view at source ↗
Figures from the paper (18 more)
Figure 4
Figure 4. Figure 4: Illustration of Base Shape Projection in 2D. Point x in Eu￾clidean space is parameterized as the signed distance s(x) and the projected footpoint xc. 3.1.3 Differentiable Projection Layer The step of ray casting makes the projected coordinates xc non-differentiable wit…
Figure 5
Figure 5. Figure 5: Shading Decomposition. Our model predicts the fine normal nf and decomposes the radiance into diffuse and specular components [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: Patch Quilting. The overlapping regions F ov 1 and F ov 2 are stitched together as F ov 1 + F ov 2 based on the minimum cost path along the error map e. The cutting path is marked in white. texture. To determine the optimal vertical cut from bottom to top, we leverage …
Figure 6
Figure 6. Figure 6: Texture Patch Extraction. We extract implicit texture patches by sampling them on the base shape, where latent features are defined. 3.2.2 Patch-based Synthesis We synthesize textures of arbitrary sizes based on the sam￾pled exemplars using patch matching and quilting …
Figure 8
Figure 8. Figure 8: Synthesis on Arbitrary Surfaces. The patch-matching synthesis algorithm of our NeRF-based texture can also be extended to arbitrary surfaces. Synthesized implicit texture is passed to the Rendering Module (RM) to produce final images of textured surface. on a high reso…
Figure 9
Figure 9. Figure 9: Feature Fetching. The synthesized features and mask are ob￾tained by querying the synthesizing UV feature map with UV coordinates of the template patch. UV grids template interpolate fill in features UV feature map [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]
Figure 10
Figure 10. Figure 10: Patch Pasting. The features of the UV grid are calculated via barycentric weighting on the template patch and the synthesized results are filled in the UV feature map. 3.2.4 Latent Feature Clustering Ideally, the metric of latent space should be consistent with that o…
Figure 11
Figure 11. Figure 11: Texture Synthesis and Applications. We show the synthesized textures of durian, tree bark, woven basket, leaves, and flowers. The textures are also applied to different shapes. The last example is constrained synthesis guided by the text image. Scenes Views Tomato Flo…
Figure 12
Figure 12. Figure 12: Texture Synthesis on Curved Surfaces. Our synthesis algorithm can be extended to arbitrary curved surfaces, considering the continuity on the surface instead of UV space [PITH_FULL_IMAGE:figures/full_fig_p007_12.png]
Figure 13
Figure 13. Figure 13: Synthesis of View-Dependent and Meso-Structure Textures. Our NeRF-based approach synthesizes textures that capture both the view-dependent appearance (in the first two rows) and meso-structure details (in the last two rows) with accuracy and fidelity. Our method prese…
Figure 14
Figure 14. Figure 14: Qualitative Comparison of View Synthesis Results. Note that our method supports texture capture, synthesis and application while visually close to the state of the arts [PITH_FULL_IMAGE:figures/full_fig_p009_14.png]
Figure 15
Figure 15. Figure 15: Qualitative Comparison with 2D Textures. We show the rendering results of our synthesized textures and 2D textures. Our rep￾resentation of maintains realism even at high-elevation viewing angles [PITH_FULL_IMAGE:figures/full_fig_p009_15.png]
Figure 17
Figure 17. Figure 17: Impact of Clustering Constraint. With the clustering loss (w/ CL), latent features are constrained to cluster, which reduces the distance of latent features corresponding to similar textures and further reduces artifacts in synthesized results. 4.6 Ablation on Patch R…
Figure 18
Figure 18. Figure 18: Ablation on Patch Resolution. A resolution of 128×128 is sufficient to represent a patch area of the required size. 4.7 Ablation on the Number of Training Views NeRF is built on the principle that when light rays intersect a surface at the same position and from the s…
Figure 19
Figure 19. Figure 19: Ablation Study on the Number of Training Views. Using fewer training views results in ambiguity in geometry and appearance. Increasing the number of training views improves the quality of rendered textures. 4.8 Rendering Speed Analysis Our approach utilizes hash grids…
Figure 21
Figure 21. Figure 21: Challenges in Limited Coarse Extents. The repeated struc￾tures of the truss pattern are distributed on a long but narrow coarse surface, which makes sampled patches too small to synthesize high￾quality textures. The roof composed of trusses, on the contrary, has enoug…
Figure 20
Figure 20. Figure 20: Challenges on Texture Capture. Our approach fails to reconstruct and capture textures on shapes with complex coarse geometry due to the difficulty in base shape estimation and patch sampling on regions with limited spatial extents of the base surface. On the contrary,…
Figure 22
Figure 22. Figure 22: Challenges on Texture Synthesis. Our synthesis approach based on patch matching struggles to exactly preserve the continuity of highly structured textures (first row) requiring strict matching with a few captured exemplars. Our method is also agnostic to the semantic …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

85 extracted references · 73 canonical work pages

  1. [1]

    NeRF-Tex: Neural reflectance field textures,

    H. Baatz, J. Granskog, M. Papas, F. Rousselle, and J. Nov ´ak, “NeRF-Tex: Neural reflectance field textures,” in Comput. Graph. Forum, no. 6, 2022, pp. 287–301

  2. [2]

    Rendering neural materials on curved surfaces,

    A. Kuznetsov, X. Wang, K. Mullia, F. Luan, Z. Xu, M. Hasan, and R. Ramamoorthi, “Rendering neural materials on curved surfaces,” in ACM SIGGRAPH 2022 Conference Proceedings , 2022, pp. 1–9

  3. [3]

    NeRF: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P . P . Srinivasan, M. Tancik, J. T. Barron, R. Ra- mamoorthi, and R. Ng, “NeRF: Representing scenes as neural radiance fields for view synthesis,” Communications of the ACM , vol. 65, no. 1, pp. 99–106, 2021

  4. [4]

    NeuMesh: Learning disentangled neural mesh-based implicit field for geometry and texture editing,

    Bao and Yang, Z. Junyi, B. Hujun, Z. Yinda, C. Zhaopeng, and Z. Guofeng, “NeuMesh: Learning disentangled neural mesh-based implicit field for geometry and texture editing,” inECCV, 2022, pp. 597–614

  5. [5]

    NeuTex: Neural texture mapping for volumetric neural rendering,

    F. Xiang, Z. Xu, M. Hasan, Y. Hold-Geoffroy, K. Sunkavalli, and H. Su, “NeuTex: Neural texture mapping for volumetric neural rendering,” in CVPR, 2021, pp. 7119–7128

  6. [6]

    Instant neural graphics primitives with a multiresolution hash encoding,

    T. M ¨uller, A. Evans, C. Schied, and A. Keller, “Instant neural graphics primitives with a multiresolution hash encoding,” ACM T rans. Graph., vol. 41, no. 4, pp. 1–15, 2022

  7. [7]

    Approximate convex decom- position for 3D meshes with collision-aware concavity and tree search,

    X. Wei, M. Liu, Z. Ling, and H. Su, “Approximate convex decom- position for 3D meshes with collision-aware concavity and tree search,” ACM T rans. Graph., vol. 41, no. 4, pp. 1–18, 2022

  8. [8]

    NeRF- Texture: Texture synthesis with neural radiance fields,

    Y.-H. Huang, Y.-P . Cao, Y.-K. Lai, Y. Shan, and L. Gao, “NeRF- Texture: Texture synthesis with neural radiance fields,” in ACM SIGGRAPH 2023 Conference Proceedings , 2023, pp. 1–10

Show all 85 references
  1. [9]

    Plenoxels: Radiance fields without neural net- works,

    S. Fridovich-Keil, A. Yu, M. Tancik, Q. Chen, B. Recht, and A. Kanazawa, “Plenoxels: Radiance fields without neural net- works,” in CVPR, 2022, pp. 5501–5510

  2. [10]

    Relu fields: The little non-linearity that could,

    A. Karnewar, T. Ritschel, O. Wang, and N. Mitra, “Relu fields: The little non-linearity that could,” inACM SIGGRAPH 2022 Conference Proceedings, 2022, pp. 1–9

  3. [11]

    NeRF++: Analyzing and improving neural radiance fields,

    K. Zhang, G. Riegler, N. Snavely, and V . Koltun, “NeRF++: Analyzing and improving neural radiance fields,” arXiv preprint arXiv:2010.07492, 2020

  4. [12]

    Mip-NeRF 360: Unbounded anti-aliased neural radiance fields,

    J. T. Barron, B. Mildenhall, D. Verbin, P . P . Srinivasan, and P . Hed- man, “Mip-NeRF 360: Unbounded anti-aliased neural radiance fields,” in CVPR, 2022, pp. 5470–5479

  5. [13]

    Block-NeRF: Scalable large scene neural view synthesis,

    M. Tancik, V . Casser, X. Yan, S. Pradhan, B. Mildenhall, P . P . Srini- vasan, J. T. Barron, and H. Kretzschmar, “Block-NeRF: Scalable large scene neural view synthesis,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 8248–8258

  6. [14]

    SurfelNeRF: Neural surfel radiance fields for online photorealistic reconstruction of indoor scenes,

    Y. Gao, Y.-P . Cao, and Y. Shan, “SurfelNeRF: Neural surfel radiance fields for online photorealistic reconstruction of indoor scenes,” arXiv preprint arXiv:2304.08971 , 2023

  7. [15]

    DeVRF: Fast deformable voxel radiance fields for dynamic scenes,

    J.-W. Liu, Y.-P . Cao, W. Mao, W. Zhang, D. J. Zhang, J. Keppo, Y. Shan, X. Qie, and M. Z. Shou, “DeVRF: Fast deformable voxel radiance fields for dynamic scenes,” in NeurIPS, 2022, pp. 36 762– 36 775

  8. [16]

    NeuPhysics: Editable neural geometry and physics from monocular videos,

    Y.-L. Qiao, A. Gao, and M. Lin, “NeuPhysics: Editable neural geometry and physics from monocular videos,” in NeurIPS, 2022, pp. 12 841–12 854

  9. [17]

    NeRFPlayer: A streamable dynamic scene representa- tion with decomposed neural radiance fields,

    L. Song, A. Chen, Z. Li, Z. Chen, L. Chen, J. Yuan, Y. Xu, and A. Geiger, “NeRFPlayer: A streamable dynamic scene representa- tion with decomposed neural radiance fields,”IEEE TVCG, vol. 29, no. 5, pp. 2732–2742, 2023

  10. [18]

    NeRD: Neural reflectance decomposition from image collec- tions,

    M. Boss, R. Braun, V . Jampani, J. T. Barron, C. Liu, and H. Lensch, “NeRD: Neural reflectance decomposition from image collec- tions,” in ICCV, 2021, pp. 12 684–12 694

  11. [19]

    NeRV: Neural reflectance and visibility fields for relighting and view synthesis,

    P . P . Srinivasan, B. Deng, X. Zhang, M. Tancik, B. Mildenhall, and J. T. Barron, “NeRV: Neural reflectance and visibility fields for relighting and view synthesis,” in CVPR, 2021, pp. 7495–7504

  12. [20]

    NeROIC: Neural rendering of objects from online image collections,

    Z. Kuang, K. Olszewski, M. Chai, Z. Huang, P . Achlioptas, and S. Tulyakov, “NeROIC: Neural rendering of objects from online image collections,” ACM T rans. Graph. , vol. 41, no. 4, pp. 1–12, 2022

  13. [21]

    Extracting triangular 3D models, materi- als, and lighting from images,

    J. Munkberg, J. Hasselgren, T. Shen, J. Gao, W. Chen, A. Evans, T. M¨uller, and S. Fidler, “Extracting triangular 3D models, materi- als, and lighting from images,” in CVPR, 2022, pp. 8280–8290

  14. [22]

    ARF: Artistic radiance fields,

    K. Zhang, N. Kolkin, S. Bi, F. Luan, Z. Xu, E. Shechtman, and N. Snavely, “ARF: Artistic radiance fields,” in ECCV, 2022, pp. 717–733

  15. [23]

    Stylized- NeRF: consistent 3D scene stylization as stylized NeRF via 2D-3D mutual learning,

    Y.-H. Huang, Y. He, Y.-J. Yuan, Y.-K. Lai, and L. Gao, “Stylized- NeRF: consistent 3D scene stylization as stylized NeRF via 2D-3D mutual learning,” in CVPR, 2022, pp. 18 342–18 352

  16. [24]

    Unified implicit neural stylization,

    Z. Fan, Y. Jiang, P . Wang, X. Gong, D. Xu, and Z. Wang, “Unified implicit neural stylization,” in ECCV, 2022, pp. 636–654

  17. [25]

    Geometry- consistent neural shape representation with implicit displacement fields,

    Y. Wang, L. Rahmann, and O. Sorkine-hornung, “Geometry- consistent neural shape representation with implicit displacement fields,” in ICLR, 2021

  18. [26]

    NeRF synthesis with shading guidance,

    C. Li, Y. Xin, G. Liu, X. Zeng, and L. Liu, “NeRF synthesis with shading guidance,” arXiv preprint arXiv:2306.11556 , 2023

  19. [27]

    Direct voxel grid optimiza- tion: Super-fast convergence for radiance fields reconstruction,

    C. Sun, M. Sun, and H.-T. Chen, “Direct voxel grid optimiza- tion: Super-fast convergence for radiance fields reconstruction,” in CVPR, 2022, pp. 5459–5469

  20. [28]

    State of the art in example-based texture synthesis,

    L. Wei, S. Lefebvre, V . Kwatra, and G. Turk, “State of the art in example-based texture synthesis,” in Eurographics, 2009, pp. 93– 117

  21. [29]

    Texture synthesis by non-parametric sampling,

    A. A. Efros and T. K. Leung, “Texture synthesis by non-parametric sampling,” in ICCV, 1999, pp. 1033–1038

  22. [30]

    Fast texture synthesis using tree- structured vector quantization,

    L.-Y. Wei and M. Levoy, “Fast texture synthesis using tree- structured vector quantization,” in Proceedings of the 27th Annual Conference on Computer Graphics and Interactive T echniques, SIG- GRAPH, 2000, p. 479–488

  23. [31]

    Real-time texture synthesis by patch-based sampling,

    L. Liang, C. Liu, Y.-Q. Xu, B. Guo, and H.-Y. Shum, “Real-time texture synthesis by patch-based sampling,” ACM T rans. Graph., vol. 20, no. 3, p. 127–150, 2001

  24. [32]

    Image quilting for texture syn- thesis and transfer,

    A. A. Efros and W. T. Freeman, “Image quilting for texture syn- thesis and transfer,” in Proceedings of the 28th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH , 2001, p. 341–346

  25. [33]

    Graphcut textures: Image and video synthesis using graph cuts,

    V . Kwatra, A. Sch ¨odl, I. Essa, G. Turk, and A. Bobick, “Graphcut textures: Image and video synthesis using graph cuts,”ACM T rans. Graph., vol. 22, no. 3, p. 277–286, 2003

  26. [34]

    PatchNet: A patch-based image representation for interactive library-driven image editing,

    S.-M. Hu, F.-L. Zhang, M. Wang, R. R. Martin, and J. Wang, “PatchNet: A patch-based image representation for interactive library-driven image editing,” ACM T rans. Graph., vol. 32, no. 6, pp. 1–12, 2013

  27. [35]

    Texture opti- mization for example-based synthesis,

    V . Kwatra, I. Essa, A. Bobick, and N. Kwatra, “Texture opti- mization for example-based synthesis,” in Proceedings of the 32th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH, 2005, p. 795–802

  28. [36]

    Texture synthesis using convolutional neural networks,

    L. A. Gatys, A. S. Ecker, and M. Bethge, “Texture synthesis using convolutional neural networks,” in NeurIPS, 2015, pp. 262–270

  29. [37]

    Very deep convolutional net- works for large-scale image recognition,

    K. Simonyan and A. Zisserman, “Very deep convolutional net- works for large-scale image recognition,” in ICLR, 2015

  30. [38]

    Tex- ture networks: Feed-forward synthesis of textures and stylized images,

    D. Ulyanov, V . Lebedev, A. Vedaldi, and V . S. Lempitsky, “Tex- ture networks: Feed-forward synthesis of textures and stylized images,” in ICML, vol. 48, 2016, pp. 1349–1357

  31. [39]

    Perceptual losses for real-time style transfer and super-resolution,

    J. Johnson, A. Alahi, and L. Fei-Fei, “Perceptual losses for real-time style transfer and super-resolution,” in ECCV, 2016, pp. 694–711

  32. [40]

    Texture synthesis with spatial generative adversarial networks,

    N. Jetchev, U. Bergmann, and R. Vollgraf, “Texture synthesis with spatial generative adversarial networks,” in Workshop on Adversar- ial T raining, NeurIPS, 2016

  33. [41]

    Precomputed real-time texture synthesis with markovian generative adversarial networks,

    C. Li and M. Wand, “Precomputed real-time texture synthesis with markovian generative adversarial networks,” in ECCV, 2016, pp. 702–716

  34. [42]

    Non-stationary texture synthesis by adversarial expansion,

    Y. Zhou, Z. Zhu, X. Bai, D. Lischinski, D. Cohen-Or, and H. Huang, “Non-stationary texture synthesis by adversarial expansion,” ACM T rans. Graph., vol. 37, no. 4, pp. 1–13, 2018

  35. [43]

    GramGAN: Deep 3d texture synthesis from 2d exemplars,

    T. Portenier, S. A. Bigdeli, and O. Goksel, “GramGAN: Deep 3d texture synthesis from 2d exemplars,” in NeurIPS, 2020, pp. 6994– 7004

  36. [44]

    Deep geomet- ric texture synthesis,

    A. Hertz, R. Hanocka, R. Giryes, and D. Cohen-Or, “Deep geomet- ric texture synthesis,” ACM T rans. Graph., vol. 39, no. 4, pp. 1–11, 2020

  37. [45]

    MeshCNN: A network with an edge,

    R. Hanocka, A. Hertz, N. Fish, R. Giryes, S. Fleishman, and D. Cohen-Or, “MeshCNN: A network with an edge,” ACM T rans. Graph., vol. 38, no. 4, pp. 1–12, 2019

  38. [46]

    Learning texture man- ifolds with the periodic spatial GAN,

    U. Bergmann, N. Jetchev, and R. Vollgraf, “Learning texture man- ifolds with the periodic spatial GAN,” in ICML, 2017, p. 469–477

  39. [47]

    Exemplar-based pattern synthesis with implicit periodic field network,

    H. Chen, J. Liu, W. Chen, S. Liu, and Y. Zhao, “Exemplar-based pattern synthesis with implicit periodic field network,” in CVPR, 2022, pp. 3708–3717

  40. [48]

    Marching cubes: A high resolu- tion 3D surface construction algorithm,

    W. E. Lorensen and H. E. Cline, “Marching cubes: A high resolu- tion 3D surface construction algorithm,” in Proceedings of the 14th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH, 1987, p. 163–169. ACCEPTED BY IEEE TRANSACTIONS ON PATTERN ANAL YSI...

  41. [49]

    Structure-from-motion revis- ited,

    J. L. Sch ¨onberger and J.-M. Frahm, “Structure-from-motion revis- ited,” in CVPR, 2016, pp. 4104–4113

  42. [50]

    Pixel- wise view selection for unstructured multi-view stereo,

    J. L. Sch ¨onberger, E. Zheng, M. Pollefeys, and J.-M. Frahm, “Pixel- wise view selection for unstructured multi-view stereo,” in ECCV, 2016, pp. 501–518

  43. [51]

    Robust watertight manifold surface generation method for ShapeNet models,

    J. Huang, H. Su, and L. Guibas, “Robust watertight manifold surface generation method for ShapeNet models,” arXiv preprint arXiv:1802.01698, 2018

  44. [52]

    Improved Laplacian smoothing of noisy surface meshes,

    J. Vollmer, R. Mencl, and H. Mueller, “Improved Laplacian smoothing of noisy surface meshes,” in Comput. Graph. Forum , vol. 18, no. 3, 1999, pp. 131–138

  45. [53]

    A two-dimensional interpolation function for irregularly-spaced data,

    D. Shepard, “A two-dimensional interpolation function for irregularly-spaced data,” in Proceedings of the 1968 23rd ACM national conference, 1968, pp. 517–524

  46. [54]

    NeRFactor: Neural factorization of shape and reflectance under an unknown illumination,

    X. Zhang, P . P . Srinivasan, B. Deng, P . Debevec, W. T. Freeman, and J. T. Barron, “NeRFactor: Neural factorization of shape and reflectance under an unknown illumination,” ACM T rans. Graph., vol. 40, no. 6, pp. 1–18, 2021

  47. [55]

    BARF: Bundle- adjusting neural radiance fields,

    C.-H. Lin, W.-C. Ma, A. Torralba, and S. Lucey, “BARF: Bundle- adjusting neural radiance fields,” in ICCV, 2021, pp. 5741–5751

  48. [56]

    NeRF−−: Neural radiance fields without known camera parameters,

    Z. Wang, S. Wu, W. Xie, M. Chen, and V . A. Prisacariu, “NeRF−−: Neural radiance fields without known camera parameters,” arXiv preprint arXiv:2102.07064, 2021

  49. [57]

    Tiny CUDA neural network framework,

    T. M ¨uller, “Tiny CUDA neural network framework,” 2021, https://github.com/nvlabs/tiny-cuda-nn

  50. [58]

    Fourier features let networks learn high frequency functions in low dimensional domains,

    M. Tancik, P . P . Srinivasan, B. Mildenhall, S. Fridovich-Keil, N. Raghavan, U. Singhal, R. Ramamoorthi, J. T. Barron, and R. Ng, “Fourier features let networks learn high frequency functions in low dimensional domains,” in NeurIPS, 2020, pp. 7537–7547

  51. [59]

    A signal-processing frame- work for inverse rendering,

    R. Ramamoorthi and P . Hanrahan, “A signal-processing frame- work for inverse rendering,” in Proceedings of the 28th Annual Con- ference on Computer Graphics and Interactive T echniques, SIGGRAPH , 2001, p. 117–128

  52. [60]

    Illumination for computer generated pictures,

    B. T. Phong, “Illumination for computer generated pictures,” Communications of the ACM , vol. 18, no. 6, pp. 311–317, 1975

  53. [61]

    Stochastic sampling in computer graphics,

    R. L. Cook, “Stochastic sampling in computer graphics,” ACM T rans. Graph., vol. 5, no. 1, p. 51–72, 1986

  54. [62]

    Texture synthesis on surfaces,

    G. Turk, “Texture synthesis on surfaces,” in Proceedings of the 28th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH, 2001, p. 347–354

  55. [63]

    Texture synthesis over arbitrary manifold surfaces,

    L.-Y. Wei and M. Levoy, “Texture synthesis over arbitrary manifold surfaces,” in Proceedings of the 28th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH, 2001, p. 355–360

  56. [64]

    Lapped textures,

    E. Praun, A. Finkelstein, and H. Hoppe, “Lapped textures,” in Proceedings of the 27th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH, 2000, p. 465–470

  57. [65]

    Hierarchical pattern map- ping,

    C. Soler, M.-P . Cani, and A. Angelidis, “Hierarchical pattern map- ping,” in Proceedings of the 29th Annual Conference on Computer Graphics and Interactive T echniques, SIGGRAPH, 2002, p. 673–680

  58. [66]

    Least squares con- formal maps for automatic texture atlas generation,

    B. L ´evy, S. Petitjean, N. Ray, and J. Maillot, “Least squares con- formal maps for automatic texture atlas generation,” ACM T rans. Graph., vol. 21, no. 3, p. 362–371, 2002

  59. [67]

    Texture mapping progressive meshes,

    P . V . Sander, J. Snyder, S. J. Gortler, and H. Hoppe, “Texture mapping progressive meshes,” in Proceedings of the 28th Annual Conference on Computer Graphics and Interactive T echniques, SIG- GRAPH, 2001, pp. 409–416

  60. [68]

    Vector field design on surfaces,

    E. Zhang, K. Mischaikow, and G. Turk, “Vector field design on surfaces,” ACM T rans. Graph., vol. 25, no. 4, p. 1294–1326, 2006

  61. [69]

    The vector heat method,

    N. Sharp, Y. Soliman, and K. Crane, “The vector heat method,” ACM T rans. Graph., vol. 38, no. 3, pp. 1–19, 2019

  62. [70]

    Unsupervised deep embed- ding for clustering analysis,

    J. Xie, R. Girshick, and A. Farhadi, “Unsupervised deep embed- ding for clustering analysis,” in ICML, 2016, pp. 478–487

  63. [71]

    Visualizing data using t-SNE,

    L. Van der Maaten and G. Hinton, “Visualizing data using t-SNE,” Journal of Machine Learning Research , vol. 9, pp. 2579–2605, 2008

  64. [72]

    On information and sufficiency,

    S. Kullback and R. A. Leibler, “On information and sufficiency,” The annals of mathematical statistics , vol. 22, no. 1, pp. 79–86, 1951

  65. [73]

    Adam: A method for stochastic optimiza- tion,

    D. P . Kingma and J. Ba, “Adam: A method for stochastic optimiza- tion,” in ICLR, 2015

  66. [74]

    Ref-nerf: Structured view-dependent appearance for neural radiance fields,

    D. Verbin, P . Hedman, B. Mildenhall, T. Zickler, J. T. Barron, and P . P . Srinivasan, “Ref-nerf: Structured view-dependent appearance for neural radiance fields,” in CVPR, 2022, pp. 5481–5490

  67. [75]

    Martin newell’s original teapot,

    A. Torrence, “Martin newell’s original teapot,” in ACM SIG- GRAPH, 2006, pp. 29–es

  68. [76]

    Robust fairing via confor- mal curvature flow,

    K. Crane, U. Pinkall, and P . Schr ¨oder, “Robust fairing via confor- mal curvature flow,” ACM T rans. Graph., vol. 32, no. 4, pp. 1–10, 2013

  69. [77]

    Zippered polygon meshes from range images,

    G. Turk and M. Levoy, “Zippered polygon meshes from range images,” in ACM SIGGRAPH, 1994, pp. 311–318

  70. [78]

    Large-scale data for multiple-view stereopsis,

    H. Aanæs, R. R. Jensen, G. Vogiatzis, E. Tola, and A. B. Dahl, “Large-scale data for multiple-view stereopsis,” IJCV, vol. 120, no. 2, pp. 153–168, 2016

  71. [79]

    Multiview neural surface reconstruction by disentan- gling geometry and appearance,

    L. Yariv, Y. Kasten, D. Moran, M. Galun, M. Atzmon, R. Basri, and Y. Lipman, “Multiview neural surface reconstruction by disentan- gling geometry and appearance,” in NeurIPS, 2020, pp. 2492–2502

  72. [80]

    SinGAN: Learning a generative model from a single natural image,

    T. R. Shaham, T. Dekel, and T. Michaeli, “SinGAN: Learning a generative model from a single natural image,” in ICCV, 2019, pp. 4570–4580

  73. [81]

    Fast fixed-radius nearest neighbors: Interactive million-particle fluids,

    R. Hoetzlein, “Fast fixed-radius nearest neighbors: Interactive million-particle fluids,” in GPU T echnology Conference (GTC), 2014, pp. 1–25

  74. [82]

    Introduction to bounding volume hierarchies,

    H. J. Haverkort, “Introduction to bounding volume hierarchies,” in Part of the PhD Thesis, Utrecht University , 2004

  75. [83]

    The r*-tree: An efficient and robust access method for points and rectangles,

    N. Beckmann, H.-P . Kriegel, R. Schneider, and B. Seeger, “The r*-tree: An efficient and robust access method for points and rectangles,” in ACM SIGMOD, 1990, pp. 322–331

  76. [84]

    A survey on bounding volume hierarchies for ray tracing,

    D. Meister, S. Ogaki, C. Benthin, M. J. Doyle, M. Guthe, and J. Bittner, “A survey on bounding volume hierarchies for ray tracing,” in Comput. Graph. Forum , vol. 40, no. 2, 2021, pp. 683– 712

  77. [85]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P . Abbeel, “Denoising diffusion probabilistic models,” in NeurIPS, 2020, pp. 6840–6851

Pith tools

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