3D Surface Reconstruction with Enhanced High-Frequency Details

Cunjian Chen; Qiuhong Ke; Shikun Zhang; Yiqun Wang; Yong Li

REVIEW 4 major objections 7 minor 30 references

3D Surface Reconstruction with Enhanced High-Frequency Details

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read FreNeuS argues that guiding a NeuS-style renderer with image-gradient high-frequency maps recovers fine surface detail that random sampling smooths away, and that the guidance transfers to any NeuS-based method.

desk verdict A cheap, modular NeuS improvement with a genuinely nice transfer experiment, but the key sampling equation as written does the opposite of what the paper claims, and the mapping from image gradients to geometric detail is never validated. read the letter →

arxiv 2505.03362 v1 pith:D5FD67I2 submitted 2025-05-06 cs.CV

Shikun Zhang , Yiqun Wang , Cunjian Chen , Yong Li , Qiuhong Ke This is my paper · ORCID

classification cs.CV

keywords neuralimplicitsurfacesNeuShigh-frequencydetaildynamicraysamplinggradient-weightedloss3DsurfacereconstructionvolumerenderingChamferdistance

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

FreNeuS argues that the fine surface details neural implicit reconstruction tends to smooth away are concentrated in the high-frequency regions of the input images, and that those regions can be located cheaply from pixel gradients and fed back into the reconstruction loop. The method builds on the NeuS signed-distance-function renderer and adds two modules: dynamic ray sampling, which allocates more rays to high-frequency pixels, and a weighted color loss that emphasizes the same pixels. On the DTU benchmark the complete method lowers mean Chamfer distance from NeuS's 0.84 to 0.73, and on six NeRF-synthetic scenes it averages 1.11 against HF-NeuS's 1.12, with qualitative gains in roof textures, tooth gaps, and leg folds. The authors further claim the two modules transplant into any NeuS-based framework, demonstrating the transfer by improving HF-NeuS's DTU mean from 0.77 to 0.71.

What carries the argument

The load-bearing object is the high-frequency map. The input image is Gaussian-denoised, its horizontal and vertical gradients $G_x(x,y)$ and $G_y(x,y)$ are combined into gradient intensity $M(x,y)=\sqrt{G_x^2+G_y^2}$, and a grayscale threshold splits pixels into a high-frequency set $I_H$ and a low-frequency set $I_L$. This map drives two mechanisms: dynamic sampling, which draws $H_{\mathrm{num}}=wB$ rays from $I_H$ and $L_{\mathrm{num}}=(1-w)B$ from $I_L$ in each training step, and a high-frequency constraint, a per-pixel weight map $W_{\mathrm{High}}$ that assigns weight $a$ to high-frequency pixels and $b$ to all others inside the color loss. The total training loss combines Eikonal regularization with $\lambda$ times this weighted color loss, with $\lambda=1.2$.

What would settle it

Render a synthetic scene with a flat plane covered by a high-contrast printed texture, reconstruct it with FreNeuS and with NeuS, and compare both surfaces to the known ground-truth plane. If FreNeuS allocates extra rays and loss weight to the textured pixels yet the recovered geometry stays flat and its Chamfer distance to ground truth is no better than NeuS's, the gradient map is responding to appearance rather than surface detail.

Watch

Extended reading notes

Core claim

The central claim is that a thresholded gradient-intensity map of each input view is a reliable proxy for where 3D surface detail is located, and that using this map twice—once to reallocate ray samples and once to reweight the color reconstruction loss—recovers detail that random sampling misses without adding a second network or extra training time. FreNeuS reports mean Chamfer distances of 0.73 on the DTU benchmark (baseline NeuS: 0.84) and 1.11 on six NeRF-synthetic scenes (HF-NeuS: 1.12), along with qualitative improvements such as the sail cut-outs in the Ship scene and the inner circle of the ball in Material. The paper also claims the two modules are framework-agnostic within the NeuS family, and demonstrates this by adding them to HF-NeuS, whose mean DTU error drops from 0.77 to 0.71.

Load-bearing premise

The whole pipeline rests on the assumption that pixels with large image-gradient intensity mark the locations of true 3D surface detail, even though color and illumination changes can also produce strong gradients on geometrically flat regions.

Editorial extensions

If this is right

Relying on image gradients rather than extra network capacity means fine details can be recovered at essentially NeuS's training cost, about 9 hours on an RTX A6000 and roughly 30 MB per scene.
FreNeuS reaches an average Chamfer distance of 0.73 on DTU, below NeuS (0.84), HF-NeuS (0.77), and 2DGS (0.80) on the same scenes.
On six NeRF-synthetic scenes the method averages 1.11, slightly below HF-NeuS's 1.12, with the largest gains in the Ship and Material scenes.
Adding the two modules to HF-NeuS lowers its DTU average from 0.77 to 0.71, supporting the claim that the guidance transfers to any NeuS-based framework.
Against LoD-NeuS's global enhanced sampling, FreNeuS matches its average error (0.73 vs 0.72) while using about 8.7 GB GPU memory instead of 13 GB.

Reading between the lines

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

Applied to depth or normal maps instead of RGB color, the same gradient-guided reweighting could separate geometric detail from painted texture, a variant the paper does not test.
The gradient map could also steer densification or loss weighting in Gaussian-splatting reconstruction, since the paper cites over-reconstruction in high-variance image regions as that family's weakness.
Using a per-scene adaptive threshold instead of a fixed grayscale threshold would stabilize the high-frequency set across datasets with different contrast, which the paper leaves unspecified.

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.

Desk Editor's Note

read the letter

Quick read on FreNeuS. The new thing is simple: take an image-gradient-based high-frequency mask and use it twice, once to allocate rays between high- and low-frequency pixels and once to weight the color loss. The combination inside NeuS is not in the cited prior work, and the generalization experiment on HF-NeuS is a genuinely nice touch — both modules improve a second NeuS-based method, which supports the modularity claim better than most papers of this kind. The gains are real but modest (DTU mean 0.73 vs 0.84 for NeuS; NeRF-synthetic 1.11 vs 1.12 for HF-NeuS), and the ablation shows both modules earn their keep.

That said, the most load-bearing equation is wrong as written. Eq. (3) sets w = IH/IL and Hnum = w*B. In any normal image IH << IL, so w < 1 and the high-frequency set gets fewer than half the rays. That is the opposite of the stated objective, which is to increase sampling in detailed regions. The intended allocation is unrecoverable from the paper; either the ratio is inverted or the formula is restructured.

The bigger conceptual soft spot is the unvalidated equation 'image gradient equals geometric detail'. Section 3.1 groups depth, color, and illumination changes into one class, and the same mask drives both sampling and loss weighting. On a geometrically flat but texture-rich surface, extra rays and extra loss weight will push the SDF to encode appearance as shape. Neither benchmark separates texture from geometry, so the reported Chamfer gains are consistent with better photometric fitting rather than better geometry. The Mic result (2.7 vs 0.72 for HF-NeuS) shows the mechanism can hurt, and the authors' explanation is speculative. This is the central claim, and no experiment isolates it.

Minor but real: the high-frequency threshold, the a and b weights, and the Gaussian parameters are never given, so reproduction is impossible; six of eight NeRF-synthetic scenes are used without explanation; no code is released; and the reference list has formatting sloppiness. The circularity concern one might raise about a training heuristic does not apply here — nothing is fitted to the observed Chamfer values, and the paper measures rather than predicts them.

Bottom line: a serious referee can get value from this. The recipe is cheap and modular, and the HF-NeuS transfer experiment is a good sign. But the authors need to fix Eq. (3), report the missing hyperparameters, and add an experiment that decouples texture from geometry before the phrase 'reconstructs fine surface details' means what they want it to mean.

Referee Report

4 major / 7 minor

Summary. This paper proposes FreNeuS, an extension of the NeuS neural implicit surface reconstruction framework. The method computes a high-frequency map per input image from Gaussian-filtered gradient magnitude (Section 3.1), partitions pixels into high-frequency (IH) and low-frequency (IL) sets, and uses this partition in two ways: to allocate ray samples between the two sets (Eq. (3)) and to weight the photometric color loss (Eqs. (4)-(5)). Experiments on 15 DTU scenes and six NeRF-synthetic scenes report Chamfer-distance improvements over NeuS (0.84 -> 0.73 on DTU) and over HF-NeuS (1.12 -> 1.11 on synthetic), together with ablations (Table 3), a generalization experiment on HF-NeuS (Table 5), and runtime/memory comparisons (Tables 4 and 6).

Significance. The proposed modules are cheap, modular, and are demonstrated to transfer to HF-NeuS, which makes the idea potentially useful to the community if the mechanism works as described. The paper deserves credit for standard benchmark evaluation, an honest ablation study, a generalization experiment, and for reporting the negative Mic result rather than omitting it; the claims are falsifiable and the computational overhead is small (roughly 3 minutes of detection per scene). I concur with the review note that the circularity concern does not land: the high-frequency map is computed from the input images, and Chamfer distance is measured post hoc. However, the significance of the reported gains is limited by three load-bearing problems: the sampling equation as written does not implement the stated oversampling objective; the assumed equivalence between image gradients and geometric detail is unvalidated and can bias the SDF toward appearance; and key hyperparameters (detection threshold, a, b) are unreported, so the method cannot be checked or reproduced. The synthetic-benchmark headline improvement is 0.01 and is driven by one outlier scene.

major comments (4)

[3.2, Eq. (3)] The sampling allocation as written cannot deliver the stated objective. With w = IH/IL (a ratio of pixel counts), the fraction of the batch B allocated to high-frequency pixels is Hnum/B = w, while the fraction of the image that is high-frequency is IH/(IH+IL); the per-pixel oversampling factor relative to uniform random sampling is (IH+IL)/IL, which is close to 1 precisely when high-frequency pixels are rare (the regime the paper says motivates the method) and which grows only as the high-frequency fraction approaches 1/2. In the opposite regime, IH >= IL gives w >= 1 and Lnum = (1-w)*B <= 0, which is not a valid allocation. For example, a scene with 5% high-frequency pixels (w = 0.053, B = 512) yields Hnum of about 27 rays, compared with about 26 rays under uniform sampling, so the mechanism is nearly a no-op in the intended setting and cannot explain the +sampling gain in Table 3 (0.84 -> 0.75). Please state the exact allocation used in the experiments (for instance, an explicit oversampling factor beyond the population ratio) and correct Eq. (3) and the surrounding text accordingly.
[3.1, Eqs. (2)-(5)] The load-bearing assumption that thresholded image-gradient magnitude marks geometric surface detail is never validated. Section 3.1 bundles depth value changes, color changes, and illumination changes into a single high-frequency class, so albedo texture on a geometrically flat surface is labeled high-frequency and receives both more rays (Eq. (3)) and a larger color-loss weight (Eqs. (4)-(5)). Because these are the only two mechanisms added to NeuS, a texture/geometry mismatch directly biases the learned SDF toward encoding appearance as geometry. The paper's own Mic result (Table 2: CD 2.70 vs. 0.72 for HF-NeuS) is consistent with this risk, and the offered explanation (lack of high-frequency lines in the stent portion) is not obviously supported by the scene content. No experiment separates texture from shape; a concrete test would be a synthetic scene with strong albedo texture on a flat surface, or a shape with geometric detail but uniform albedo. The authors should add such an experiment or otherwise show that the high-frequency map tracks geometry rather than appearance.
[3.1 and 3.3] Reproducibility: key parameters and definitions are missing. Section 3.1 says boundary tracing is performed based on the grayscale threshold, but the threshold value is never reported; the threshold of 25 mentioned in Section 4 is the surface-extraction threshold for the reconstructed meshes, not the high-frequency detection threshold. The loss weights a and b in Eq. (4) are likewise never specified, so the weighted color loss in Eq. (5) is underspecified, and it is not stated whether a > 1 > b or some other relation is intended. In addition, Eq. (2) is circular as written (the condition defining IH refers to P in IH), and the relationship between the continuous gradient magnitude M(x,y) of Eq. (1) and the binary partition of Eq. (2) is unclear. Without these values the method cannot be reproduced, and the sensitivity of Tables 1-3 to these parameters is not assessed.
[Tables 1-3] Tables 1-3 report single runs with no variance or multiple seeds, and the headline synthetic-benchmark improvement is fragile. In Table 2, the mean Chamfer-distance improvement over HF-NeuS is 1.11 vs. 1.12 (about 1%) and is driven by the Materials scene (1.08 -> 0.13), while FreNeuS is substantially worse on Lego (1.30 vs. 0.94) and Mic (2.70 vs. 0.72); no discussion of this outlier is given. In the DTU ablation (Table 3), the +sampling and +Fre-constraints rows each produce the same mean (0.75) with different per-scene patterns, which is consistent with run-to-run variation. The authors should report standard deviations or multiple seeds and discuss the Materials result before claiming consistent improvement over HF-NeuS.

minor comments (7)

[Eq. (6)] The Eikonal loss has mismatched parentheses; it should read (1/|I|) * sum over i of (||grad S(x_i,y_i,z_i)|| - 1).
[Table 6] The FreNeuS training-time entry reads '9' without the 'h' unit used elsewhere in the table.
[References and Table 2] References [25] and [29] both cite the 2DGS paper, and the baseline list in Section 4 labels [28] as VolSDF although VolSDF is [4]; the label 'VOLSD' in Table 2 is a typo.
[4.2.3] The text claims that neither proposed module adds extra computational time and then states that high-frequency detection takes about 3 minutes per scene; these statements should be reconciled.
[Figure 3 caption] The caption says 'Second to fifth columns: NeuS, HF-NeuS, 3DGS, 2DGS, and OURS' but lists five methods for four columns.
[Section 1] The description of Mip-NeRF and Tri-MipRF as expanding the sampling area around each ray is inaccurate; these methods integrate over cone or tri-mip regions for anti-aliasing rather than widening per-ray point sampling.
[Section 3.1] The text mentions 'the first Gaussian filter G(x,y)', implying a second filter that is never introduced; the filter size and sigma are also unreported.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the method is a supervised training heuristic whose reported Chamfer improvements are measured against external geometry, not derived from the high-frequency map by construction.

full rationale

The paper's central claim is that guiding ray sampling and color-loss weighting with a thresholded gradient map improves NeuS-style surface reconstruction. This is an empirical training heuristic, not a derivation of the target metric. The high-frequency map is computed from the input images via Eq. (1) (gradient magnitude) and Eq. (2) (thresholding), then used in Eq. (3) to split the ray budget and in Eqs. (4)-(5) to weight the photometric loss. Nothing in these equations defines the high-frequency map in terms of Chamfer distance or the extracted SDF, and no parameter is fitted so that the reported DTU or NeRF-synthetic Chamfer improvements are forced. The baselines (NeuS, VolSDF, HF-NeuS, 3DGS, 2DGS) are external, and the comparison metrics are computed after training against ground-truth geometry, so the paper does not rename a fitted input as a prediction. The one author-overlapping citation, HF-NeuS, is used as a baseline and as a host for the generalization experiments; it is not load-bearing for the method's derivation. The conflation of color/illumination gradients with geometric detail is a genuine correctness and robustness concern, and the unspecified threshold, loss weights a and b, and sampling weight w make the mechanism hard to reproduce, but these are not circularity: the reported improvements remain externally falsifiable measurements rather than consequences of the paper's definitions. No self-definitional step, fitted-input-as-prediction, or self-citation chain is present, so the appropriate finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The method's novelty is a heuristic sampling and weighting scheme; it introduces several hand-set hyperparameters and relies on the assumption that image-space gradient magnitude indicates geometric surface detail. No new physical or geometric entities are introduced.

free parameters (4)

high-frequency loss weights a and b
Eqn. (4) assigns weights a and b to high- and low-frequency rays in the color loss; the paper never reports their values, yet they directly control the strength of the proposed constraint.
high-frequency detection threshold
Section 3.1 uses a grayscale threshold for boundary tracing to build the high-frequency map, but the threshold value is never given.
Gaussian filter parameters
Section 3.1 applies a Gaussian filter before gradient computation; kernel size and sigma are not specified, affecting which structures are classified as high-frequency.
loss weight lambda = 1.2
Section 3.4 sets lambda = 1.2 for the color loss L_Frecolor; it is chosen by hand with no sensitivity analysis or justification.

assumptions (3)

domain assumption Thresholded image-gradient magnitude is a reliable proxy for high-frequency surface geometry, including depth, color and illumination changes.
Section 3.1 identifies high-frequency points using image gradients and assumes these correspond to surface detail; texture and lighting edges may not correspond to geometric detail.
domain assumption High-frequency maps computed per input image are consistent enough across views to guide a shared SDF optimization without explicit multi-view handling.
The HF map is computed independently for each training image and used to sample rays and weight losses; view-dependent or inconsistent high-frequency signals are not addressed.
standard math The standard NeuS volume rendering and Eikonal regularization remain valid when combined with the new sampling and loss weighting.
Section 3.4 reuses NeuS's rendering loss and Eikonal loss without modification, so the method inherits the standard assumptions of neural implicit surface rendering.

how reviews work

0 comments

Cite this review

Pith. "Pith review of 3D Surface Reconstruction with Enhanced High-Frequency Details." pith.science (2026). https://pith.science/paper/D5FD67I2

@misc{pith2026250503362,
  author       = {Pith},
  title        = {Pith review of: 3D Surface Reconstruction with Enhanced High-Frequency Details},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D5FD67I2}},
  note         = {Machine review of arXiv:2505.03362}
}

read the original abstract

Neural implicit 3D reconstruction can reproduce shapes without 3D supervision, and it learns the 3D scene through volume rendering methods and neural implicit representations. Current neural surface reconstruction methods tend to randomly sample the entire image, making it difficult to learn high-frequency details on the surface, and thus the reconstruction results tend to be too smooth. We designed a method (FreNeuS) based on high-frequency information to solve the problem of insufficient surface detail. Specifically, FreNeuS uses pixel gradient changes to easily acquire high-frequency regions in an image and uses the obtained high-frequency information to guide surface detail reconstruction. High-frequency information is first used to guide the dynamic sampling of rays, applying different sampling strategies according to variations in high-frequency regions. To further enhance the focus on surface details, we have designed a high-frequency weighting method that constrains the representation of high-frequency details during the reconstruction process. Qualitative and quantitative results show that our method can reconstruct fine surface details and obtain better surface reconstruction quality compared to existing methods. In addition, our method is more applicable and can be generalized to any NeuS-based work.

Figures

Figures reproduced from arXiv: 2505.03362 by the authors.

**Figure 2.** The figure shows the obtained high-frequency image, along with the RGB images [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Visualisation comparisons on different scenes of the DTU dataset. First column: [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Qualitative evaluation on the Ship and Material scenes. First column: reference [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

**Figure 5.** Figure 5: Visualisation of ablation experiments to verify the performance of each module. [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]

**Figure 6.** Figure 6: Visual comparisons of improved HF-NeuS after integrating the two proposed [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 24 canonical work pages

[1]

Hartley, A

R. Hartley, A. Zisserman, Multiple view geometry in computer vision, Cambridge university press, 2003. 2

work page 2003
[2]

Mildenhall, P

B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoor- thi, R. Ng, Nerf: Representing scenes as neural radiance fields for view synthesis, Communications of the ACM 65 (1) (2021) 99–106. 2, 5, 10, 11, 12

work page 2021
[3]

P. Wang, L. Liu, Y. Liu, C. Theobalt, T. Komura, W. Wang, Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction, arXiv preprint arXiv:2106.10689 (2021). 2, 3, 5, 10

arXiv 2021
[4]

Yariv, J

L. Yariv, J. Gu, Y. Kasten, Y. Lipman, Volume rendering of neural implicit surfaces, Advances in Neural Information Processing Systems 34 (2021) 4805–4815. 2, 5

work page 2021
[5]

J. T. Barron, B. Mildenhall, M. Tancik, P. Hedman, R. Martin-Brualla, P. P. Srinivasan, Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 5855–5864. 2 16

work page 2021
[6]

W. Hu, Y. Wang, L. Ma, B. Yang, L. Gao, X. Liu, Y. Ma, Tri-miprf: Tri- mip representation for efficient anti-aliasing neural radiance fields, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 19774–19783. 2

work page 2023
[7]

H. Guo, S. Peng, H. Lin, Q. Wang, G. Zhang, H. Bao, X. Zhou, Neural 3d scene reconstruction with the manhattan-world assumption, in: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5511–5520. 2

work page 2022
[8]

Z. Yu, S. Peng, M. Niemeyer, T. Sattler, A. Geiger, Monosdf: Exploring monocular geometric cues for neural implicit surface reconstruction, Ad- vances in neural information processing systems 35 (2022) 25018–25032. 2

work page 2022

Show all 30 references

[9]

Kerbl, G

B. Kerbl, G. Kopanas, T. Leimk¨ uhler, G. Drettakis, 3d gaussian splat- ting for real-time radiance field rendering, ACM Transactions on Graph- ics 42 (4) (2023) 1–14. 2, 5

2023
[10]

Zhang, F

J. Zhang, F. Zhan, M. Xu, S. Lu, E. Xing, Fregs: 3d gaussian splat- ting with progressive frequency regularization, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21424–21433. 2

2024
[11]

Broadhurst, T

A. Broadhurst, T. W. Drummond, R. Cipolla, A probabilistic framework for space carving, in: Proceedings eighth IEEE international conference on computer vision. ICCV 2001, Vol. 1, IEEE, 2001, pp. 388–393. 5

2001
[12]

J. S. De Bonet, P. Viola, Poxels: Probabilistic voxelized volume re- construction, in: Proceedings of International Conference on Computer Vision (ICCV), Vol. 2, Citeseer, 1999, p. 2. 5

1999
[13]

Izadi, D

S. Izadi, D. Kim, O. Hilliges, D. Molyneaux, R. Newcombe, P. Kohli, J. Shotton, S. Hodges, D. Freeman, A. Davison, et al., Kinectfusion: real-time 3d reconstruction and interaction using a moving depth cam- era, in: Proceedings of the 24th annual ACM symposium on User inter- f...

2011
[14]

K. N. Kutulakos, S. M. Seitz, A theory of shape by space carving, In- ternational journal of computer vision 38 (2000) 199–218. 5 17

2000
[15]

Nießner, M

M. Nießner, M. Zollh¨ ofer, S. Izadi, M. Stamminger, Real-time 3d recon- struction at scale using voxel hashing, ACM Transactions on Graphics (ToG) 32 (6) (2013) 1–11. 5

2013
[16]

S. M. Seitz, C. R. Dyer, Photorealistic scene reconstruction by voxel coloring, International journal of computer vision 35 (1999) 151–173. 5

1999
[17]

Barnes, E

C. Barnes, E. Shechtman, A. Finkelstein, D. B. Goldman, Patchmatch: A randomized correspondence algorithm for structural image editing, ACM Trans. Graph. 28 (3) (2009) 24. 5

2009
[18]

Galliani, K

S. Galliani, K. Lasinger, K. Schindler, Gipuma: Massively parallel multi- view stereo reconstruction, Publikationen der Deutschen Gesellschaft f¨ ur Photogrammetrie, Fernerkundung und Geoinformation e. V 25 (361-
[19]

J. L. Schonberger, J.-M. Frahm, Structure-from-motion revisited, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 4104–4113. 5

2016
[20]

J. L. Sch¨ onberger, E. Zheng, J.-M. Frahm, M. Pollefeys, Pixelwise view selection for unstructured multi-view stereo, in: Computer Vision– ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III 14, Springer, 2016, pp. 501–

2016
[21]

Y. Wang, I. Skorokhodov, P. Wonka, Hf-neus: Improved surface recon- struction using high-frequency details, Advances in Neural Information Processing Systems 35 (2022) 1966–1978. 5, 10

2022
[22]

Snavely, S

N. Snavely, S. M. Seitz, R. Szeliski, Photo tourism: exploring photo collections in 3d, in: ACM siggraph 2006 papers, 2006, pp. 835–846. 5

2006
[23]

Jiang, J

Y. Jiang, J. Tu, Y. Liu, X. Gao, X. Long, W. Wang, Y. Ma, Gaus- sianshader: 3d gaussian splatting with shading functions for reflective surfaces, arXiv preprint arXiv:2311.17977 (2023). 6

2023 arXiv
[24]

J. Gao, C. Gu, Y. Lin, H. Zhu, X. Cao, L. Zhang, Y. Yao, Relightable 3d gaussian: Real-time point cloud relighting with brdf decomposition and ray tracing, arXiv preprint arXiv:2311.16043 (2023). 6 18

2023 arXiv
[25]

Huang, Z

B. Huang, Z. Yu, A. Chen, A. Geiger, S. Gao, 2d gaussian splatting for geometrically accurate radiance fields, arXiv preprint arXiv:2403.17888 (2024). 6

2024 arXiv
[26]

Gropp, L

A. Gropp, L. Yariv, N. Haim, M. Atzmon, Y. Lipman, Implicit geomet- ric regularization for learning shapes, arXiv preprint arXiv:2002.10099 (2020). 10

2020 arXiv
[27]

Jensen, A

R. Jensen, A. Dahl, G. Vogiatzis, E. Tola, H. Aanæs, Large scale multi- view stereopsis evaluation, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2014, pp. 406–413. 10

2014
[28]

Yariv, Y

L. Yariv, Y. Kasten, D. Moran, M. Galun, M. Atzmon, B. Ronen, Y. Lip- man, Multiview neural surface reconstruction by disentangling geometry and appearance, Advances in Neural Information Processing Systems 33 (2020) 2492–2502. 10

2020
[29]

Huang, Z

B. Huang, Z. Yu, A. Chen, A. Geiger, S. Gao, 2d gaussian splatting for geometrically accurate radiance fields, in: SIGGRAPH 2024 Conference Papers, Association for Computing Machinery, 2024. doi:10.1145/ 3641519.3657428. 10

2024
[30]

Zhuang, Q

Y. Zhuang, Q. Zhang, Y. Feng, H. Zhu, Y. Yao, X. Li, Y.-P. Cao, Y. Shan, X. Cao, Anti-aliased neural implicit surfaces with encoding level of detail, in: SIGGRAPH Asia 2023 Conference Papers, 2023, pp. 1–10. 14 19

2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Hartley, A

R. Hartley, A. Zisserman, Multiple view geometry in computer vision, Cambridge university press, 2003. 2

work page 2003

[2] [2]

Mildenhall, P

B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoor- thi, R. Ng, Nerf: Representing scenes as neural radiance fields for view synthesis, Communications of the ACM 65 (1) (2021) 99–106. 2, 5, 10, 11, 12

work page 2021

[3] [3]

P. Wang, L. Liu, Y. Liu, C. Theobalt, T. Komura, W. Wang, Neus: Learning neural implicit surfaces by volume rendering for multi-view reconstruction, arXiv preprint arXiv:2106.10689 (2021). 2, 3, 5, 10

arXiv 2021

[4] [4]

Yariv, J

L. Yariv, J. Gu, Y. Kasten, Y. Lipman, Volume rendering of neural implicit surfaces, Advances in Neural Information Processing Systems 34 (2021) 4805–4815. 2, 5

work page 2021

[5] [5]

J. T. Barron, B. Mildenhall, M. Tancik, P. Hedman, R. Martin-Brualla, P. P. Srinivasan, Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 5855–5864. 2 16

work page 2021

[6] [6]

W. Hu, Y. Wang, L. Ma, B. Yang, L. Gao, X. Liu, Y. Ma, Tri-miprf: Tri- mip representation for efficient anti-aliasing neural radiance fields, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 19774–19783. 2

work page 2023

[7] [7]

H. Guo, S. Peng, H. Lin, Q. Wang, G. Zhang, H. Bao, X. Zhou, Neural 3d scene reconstruction with the manhattan-world assumption, in: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5511–5520. 2

work page 2022

[8] [8]

Z. Yu, S. Peng, M. Niemeyer, T. Sattler, A. Geiger, Monosdf: Exploring monocular geometric cues for neural implicit surface reconstruction, Ad- vances in neural information processing systems 35 (2022) 25018–25032. 2

work page 2022