Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Sparse2DGS shows that seeding 2D Gaussian splatting with an MVS point cloud and prioritizing geometry yields accurate surfaces from three views.

desk verdict A solid, well-ablated empirical paper on sparse-view Gaussian splatting that deserves peer review, though the single-run numbers and an overstateful abstract need fixing. read the letter →

arxiv 2504.20378 v1 pith:NECX7YEV submitted 2025-04-29 cs.CV

Jiang Wu , Rui Li , Yu Zhu , Rong Guo , Jinqiu Sun , Yanning Zhang This is my paper · ORCID
classification cs.CV
keywords sparse-view3DreconstructionGaussiansplattingsurfacemulti-viewstereo2Dprimitivescross-viewconsistencygeometricregularizationDTUdataset
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

Sparse2DGS sets out to show that accurate and complete 3D surface reconstruction is possible from as few as three posed images, by seeding 2D Gaussian Splatting with a learning-based multi-view stereo (MVS) point cloud and then steering optimization toward geometry rather than appearance. The paper reports that a plain MVS-plus-2DGS combination fails to beat the MVS points alone (Chamfer distance 1.27 vs 1.26), but that adding three geometry-prioritized schemes—fixed-color MVS feature supervision, reparameterization-based disk regularization, and selective Gaussian updates from rendered depth and normals—lowers the DTU 3-view Chamfer distance to 1.13, below all compared Gaussian splatting methods. If this holds, sparse-view reconstruction becomes practical for scenes where only a handful of images exist, and the approach is fast enough (about 10 minutes) to serve as a practical test-time pipeline.

What carries the argument

The mechanism that carries the argument is a reparameterized 2D Gaussian disk: points are drawn from a standard Gaussian and transformed by $Z = p + RSZ'$, where $R$ is the disk orientation, $S$ its scale, and $p$ its position, so all three geometric properties receive gradients from a single point-based cross-view feature consistency loss. This direct Gaussian primitive regularization is backed by geometrically enhanced supervision, in which CLMVSNet FPN features and RGB colors are splatted into the renderer and held fixed, with rendered features supervised by cosine similarity, and by selective Gaussian update, in which patch-wise normalized cross-correlation over rendered depth and normal maps decides when to reproject a point and replace a primitive position. Together these convert a Gaussian primitive into a set of query points that can be checked for multi-view agreement, while keeping appearance from absorbing the sparse-view ambiguity.

What would settle it

On a sparse-view set with a large occluded region, measure per-region Chamfer error against ground truth for Sparse2DGS and for the plain 2DGS-plus-CLMVSNet baseline; if the occluded regions do not improve, or if deliberately corrupting CLMVSNet depth in those regions leaves the overall error essentially unchanged, then the geometric-prioritized schemes are not doing the corrective work the paper attributes to them.

Watch

Extended reading notes

Core claim

The central claim is that the main obstacle to sparse-view Gaussian surface reconstruction is not missing primitives but missing geometric constraint, and that MVS-derived points plus features can supply that constraint if geometry is prioritized. Sparse2DGS initializes every 2D disk primitive from back-projected CLMVSNet depth, freezes per-primitive color and MVS FPN feature values so appearance cannot overfit the few views, represents each disk's position, orientation, and scale through sampled points via $Z = p + RSZ'$ supervised by cross-view feature consistency, and replaces adaptive densification with a selective update that repositions a primitive when patch-wise NCC favors the rendered depth/normal cues. On the DTU three-view benchmark the paper reports a mean Chamfer distance of 1.13, compared with 2.81 for 2DGS, 2.82 for GOF, and 2.08 for PGSR, with training time about 10 minutes—2x faster than fine-tuning SparseNeuS and roughly 60x faster than NeuSurf.

Load-bearing premise

The load-bearing premise is that the MVS feature maps, computed once from the three input views, are locally view-consistent at sub-pixel positions for arbitrary points sampled on the Gaussian disks—points that generally do not lie exactly on surfaces—and that the MVS depths themselves remain reliable in occluded and textureless regions; the paper's own limitation section concedes that occlusion errors from MVS can propagate and hurt performance.

Editorial extensions

If this is right

  • On the three-view DTU benchmark, Sparse2DGS reports mean Chamfer distance 1.13, below 2DGS (2.81), GOF (2.82), and PGSR (2.08).
  • The method reconstructs more complete surfaces in qualitative comparisons on DTU and BlendedMVS, preserving scene structure where prior Gaussian methods produce noise and missing regions.
  • Training finishes in about 10 minutes, roughly 2x faster than SparseNeuS fine-tuning and about 60x faster than NeuSurf, making the approach practical as a test-time pipeline.
  • Ablations show that fixing color and MVS feature values helps geometry more than lowering spherical-harmonic degrees, supporting the paper's diagnosis of appearance overfitting under sparse views.
  • Replacing adaptive density control with selective Gaussian update avoids the degradation of densified primitives under sparse supervision in the paper's ablation (1.152 vs 1.234 Chamfer distance).

Reading between the lines

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

  • The same recipe—a dense geometric prior that supplies both points and features, frozen appearance, and point-based geometric regularization—could plausibly work with other priors such as monocular depth or foundation-model features, though the paper only tests CLMVSNet.
  • Because the cross-view consistency loss uses features extracted once, the method's ceiling is largely set by the discriminative power and view consistency of those features; improving or fine-tuning the feature extractor should directly lower reconstruction error, a testable prediction.
  • The selective update step is a local geometric refinement; running it more often or combining it with visibility checks might extend the method to scenes with larger occlusions, where the paper's stated limitation predicts degraded performance.
  • If the reported DTU margin is robust, the practical implication is that sparse-view surface reconstruction no longer requires per-scene NeRF fine-tuning or hours of optimization, which should make three-image 3D capture pipelines substantially cheaper.
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 / 6 minor

Summary. The paper proposes Sparse2DGS, a sparse-view surface reconstruction method built on 2D Gaussian Splatting (2DGS). It initializes the Gaussian cloud from the dense point cloud of the unsupervised MVS network CLMVSNet, then applies three geometry-prioritized enhancements: (i) fixed-color and fixed-feature supervision using MVS feature maps rendered through the Gaussian splatting pipeline (Sec. 3.2); (ii) a Direct Gaussian Primitive Regularization (DGPR) loss that reparameterizes each disk by sampling points from a standard Gaussian, transforming them by the primitive's position, rotation, and scale, and supervising the projected points with cross-view feature consistency (Sec. 3.3, Eqs. 6-8); and (iii) a Selective Gaussian Update (SGU) that compares patch-based NCC scores between warping using Gaussian primitives and warping using rendered depths/normals, then reprojects better rendered depths into updated positions (Sec. 3.4, Eq. 11). On the DTU dataset with three views (23, 24, 33), the method reports a mean Chamfer Distance of 1.13, outperforming 2DGS (2.81), GOF (2.82), and PGSR (2.08), while being competitive with NeRF-based sparse-view methods and about 2x faster than SparseNeuS. Ablations in Tables 2, 4, 5, and 6 attribute improvements to each proposed module.

Significance. If the results hold, this is a practical and timely contribution: it shows that a dense MVS point cloud combined with fixed-feature supervision and point-based disk regularization can make Gaussian splatting much more accurate for sparse-view surface reconstruction than standard SfM-initialized 2DGS, while keeping training time around ten minutes. The paper ships code, uses the official DTU evaluation protocol, and includes a controlled ablation that separates the MVS-initialization gain from the proposed losses. The main methodological novelty, DGPR, is clearly described and the cross-view consistency formulation is principled in the ideal case of perfectly view-consistent features. However, the central claim that the proposed regularization is responsible for the reported gains rests on an assumption about the local view consistency of fixed MVS features at off-surface samples, and the ablation margins are small and based on single runs, so the significance of the individual modules is not yet fully established.

major comments (3)
  1. [Tables 2, 5, and 6] The DGPR loss assumes that CLMVSNet FPN features are locally view-consistent at arbitrary points sampled on the 2D Gaussian disks, including points that do not lie on the true surface and that may project into occluded, textureless, or repetitive regions of the target view. The manuscript does not provide any validation of this assumption: there is no occlusion masking, no analysis of the fraction of off-surface samples that are actually visible in the target view, and no comparison against a baseline where the features are randomized or where only on-surface points are used. Since the paper's abstract and introduction attribute the key improvement to this geometric-prioritized regularization, this unvalidated assumption is load-bearing for the central claim. Please provide quantitative evidence of feature consistency, add a visibility mask or a robustness analysis, or ablate the loss against a random-feature version to show that the CD gain in Table 2 (row e to row f, 1.201 to 1.142) is not driven by spurious feature similarities.
  2. [Sec. 3.4, Eq. (11)] All experiments report a single run with no variance or seed information, and several ablation margins are small: Table 2 shows DGPR improving CD from 1.201 to 1.142 (0.059) and Table 5 shows SGU improving CD from 1.201 to 1.152 (0.049), while Table 6 shows the effect of K varying from 1.145 to 1.139. These differences are within the typical run-to-run variability of Gaussian splatting optimizations, so the per-module attribution is not statistically supported. Please report mean and standard deviation over at least three random seeds for the main ablations in Tables 2, 4, 5, and 6, and state whether the reported metrics are averaged or from a single representative run.
  3. [Sec. 4.3, Table 1] The selective update rule replaces a Gaussian position whenever NCC_R > NCC_G, with no absolute threshold on the NCC values. In textureless or homogeneous regions, even a tiny positive difference between two low NCC scores would trigger an update, potentially injecting noise from unreliable rendered depth into the point set. The paper does not report the distribution of NCC scores, the fraction of primitives updated per iteration, or the sensitivity of the final CD to an NCC threshold. Please provide such an analysis (e.g., a sweep over thresholds, or a report of typical NCC values for updated versus non-updated primitives) to demonstrate that the update is meaningful and not dominated by low-confidence matches.
minor comments (6)
  1. [Sec. 4.3, Table 1] Equation (12) contains a stray comma after \lambda_2 L_n; please correct to "\mathcal{L} = \mathcal{L}_{rgb} + \lambda_1 \mathcal{L}_d + \lambda_2 \mathcal{L}_n + \lambda_3 \mathcal{L}_{dr} + \lambda_4 \mathcal{L}_{fea}".
  2. [Sec. 3.2] The headline comparison in Table 1 uses different initialization strategies for the baselines (SfM points from three views for 2DGS/GOF/PGSR) versus the proposed method (CLMVSNet dense points). The controlled comparison in Table 3 (2DGS+CLMVSNet: 1.27, PGSR+CLMVSNet: 1.38, Ours: 1.13) is more informative for isolating the contribution of the proposed losses; please reference it directly in the main text near Table 1 to avoid overstating the margin against 2DGS.
  3. [Sec. 3.4] After Eq. (3), please clarify how the feature maps F_i are resized or interpolated when querying at fractional pixel coordinates \pi(p_{ij}), and whether the same FPN resolution (8 channels) is used for both the feature splatting loss and the DGPR cross-view consistency loss.
  4. [Sec. 4.4] The description of the selective update does not specify the patch size used for NCC, the number of update iterations over the 7000 training steps, or whether the update is applied to all Gaussians or only those with sufficient gradient. Please add these implementation details for reproducibility.
  5. [Sec. 4.3] Table 4's header contains a typo: "Compleness" should be "Completion".
  6. [Sec. 4.4] The claim that the method is "2x faster than the NeRF-based fine-tuning approach" is based on comparison with SparseNeuS (Table 7); consider stating this explicitly in the abstract and main text, since NeuSurf's training time is much larger but its CD is better (0.99).

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the DTU Chamfer evaluation is external, and no loss term reduces to the final metric or to the MVS initialization by construction.

full rationale

Sparse2DGS's derivation chain is self-contained with respect to the paper's central claims. The reported 1.13 Chamfer distance is measured against ground-truth DTU scans using the official evaluation code, and that ground-truth metric appears nowhere in the training losses in Eq. (12). The method does use CLMVSNet both to produce the initial point cloud and to supply fixed FPN feature maps for supervision (Eqs. 3-5), which makes the feature supervision self-supervised rather than externally grounded; however, this is not circular by construction. The optimization can move Gaussian primitives away from the initial MVS depth, and the reported 1.13 CD improves upon CLMVSNet's own 1.26 result and the plain 2DGS+CLMVSNet combination at 1.27, so the output is not forced to equal the MVS input. The cross-view consistency loss in Eq. (6) is a standard feature-metric regularization whose target is the fixed feature maps, not the final mesh or the evaluation metric. The selective Gaussian update uses the model's own rendered depth and normals as a self-correction cue, which is an internal consistency operation rather than a definition of the benchmark metric. No load-bearing self-citations or imported uniqueness theorems are present: references [16], [17], and [30] are background citations to the authors' prior MVS work and are not used to justify a prediction. The paper's acknowledged limitation that MVS errors under occlusion can propagate is a statement about error propagation, not a circular step. Therefore, no circularity is found.

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

The central claim rests on the reliability of MVS features as geometric supervision, the validity of cross-view feature consistency as a surface proxy, and the stability of the NCC-based selective update. These are empirical assumptions inherited from prior MVS and Gaussian Splatting work, plus hyperparameters tuned on DTU. The paper introduces no new physical entities.

free parameters (5)
  • lambda1 (depth-normal consistency weight) = 1000
    Set in Sec. 4.2; balances image reconstruction against the depth-normal consistency term in Eq. (12), tuned on the DTU benchmark.
  • lambda2 (depth distortion weight) = 0.05
    Set in Sec. 4.2; weights the depth distortion loss.
  • lambda3 (DGPR weight) = 1
    Set in Sec. 4.2; weights the direct Gaussian primitive regularization.
  • lambda4 (feature splatting loss weight) = 0.2
    Set in Sec. 4.2; weights the MVS feature supervision loss.
  • Number of sampled points K per disk = not explicitly stated for the main result
    Table 6 varies K from 9 to 81 with K=81 performing best; the paper does not state which K is used in Table 1, so the main numbers depend on an unspecified hyperparameter choice.
assumptions (5)
  • domain assumption 2D Gaussian primitives (Eq. 1-2) can represent the scene surface and support TSDF mesh extraction.
    Inherited from 2DGS [11]; if the disk representation cannot align with surfaces, the whole pipeline fails. This is the paper's base representation.
  • ad hoc to paper CLMVSNet features are geometrically expressive and view-consistent at sub-pixel locations (Secs. 3.2-3.3).
    Both the feature splatting loss (Eq. 5) and the cross-view consistency loss (Eq. 6) treat these features as reliable geometric supervision. The conclusion explicitly concedes that MVS errors under occlusion propagate, making this assumption load-bearing.
  • domain assumption Cross-view feature consistency is a valid proxy for surface correctness (Eq. 6).
    This is a standard MVS assumption, cited from Zhang et al. [45]; it holds when features are discriminative and views have overlap, but can be violated in textureless or repeated patterns.
  • ad hoc to paper NCC of warped patches is a reliable indicator of geometric quality for the selective update (Sec. 3.4, Eq. 11).
    The update rule moves primitives to rendered-depth reprojections whenever NCCR exceeds NCCG; no threshold, patch size, or failure analysis is provided, and NCC reliability in low-texture regions is not established.
  • standard math The reparameterization Z = p + RSZ' with S = diag(su,sv,0) provides valid gradients to position, orientation, and scale (Eqs. 7-8).
    This is the standard reparameterization trick for Gaussian distributions; the math is straightforward and correct.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views." pith.science (2026). https://pith.science/paper/NECX7YEV

@misc{pith2026250420378,
  author       = {Pith},
  title        = {Pith review of: Sparse2DGS: Geometry-Prioritized Gaussian Splatting for Surface Reconstruction from Sparse Views},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NECX7YEV}},
  note         = {Machine review of arXiv:2504.20378}
}
abstract

We present a Gaussian Splatting method for surface reconstruction using sparse input views. Previous methods relying on dense views struggle with extremely sparse Structure-from-Motion points for initialization. While learning-based Multi-view Stereo (MVS) provides dense 3D points, directly combining it with Gaussian Splatting leads to suboptimal results due to the ill-posed nature of sparse-view geometric optimization. We propose Sparse2DGS, an MVS-initialized Gaussian Splatting pipeline for complete and accurate reconstruction. Our key insight is to incorporate the geometric-prioritized enhancement schemes, allowing for direct and robust geometric learning under ill-posed conditions. Sparse2DGS outperforms existing methods by notable margins while being ${2}\times$ faster than the NeRF-based fine-tuning approach.

Figures

Figures reproduced from arXiv: 2504.20378 by the authors.

Figure 1
Figure 1. Sparse2DGS boosts the strengths of Gaussian Splatting and Multi-view Stereo in sparse-view surface reconstruction, with notable improvement over 2DGS [11], CLMVSNet [31], as well as their plain combination (left). Meanwhile, Sparse2DGS is 2× faster than NeRF-based sparse-view method [20] (middle) and leads to complete and accurate reconstruction (right). Abstract We present a Gaussian Splatting method for surface re… view at source ↗
Figure 2
Figure 2. Overview. Given sparse posed images, we first leverage the MVS points to initialize the Gaussian position p (Sec. 3.1). We then leverage the MVS-derived feature for Gaussian Splatting, with fixed feature f and color c values to conduct geometrically expressive supervision and avoid appearance overfitting (Sec. 3.2). We then optimize Gaussian primitive properties by reformulating orientation R, scale S, and position … view at source ↗
Figure 3
Figure 3. Direct Gaussian Primitive Regularization. We sample point set Z ′ ∼ N (0, I) from the standard Gaussian distribution and leverage the reparameterization technique to transform Z ′ to the point set Z on each Gaussian primitive. This process allows for representing Gaussian orientation R and scale S properties in the form of points easier for geometric supervision. The trans￾formed points Z are then projected to the s… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: DTU surface reconstruction results. Our method achieves more complete reconstructions with finer details. 4.2. Implementation Details Our method is implemented based on 2DGS [11]. For Gaus￾sian primitive initialization, we use the point cloud back￾projected depth maps …
Figure 5
Figure 5. Figure 5: BlendedMVS reconstruction results. Our method achieves more complete and detailed reconstructions. 2DGS MVS Init. Fixed Color Feat. Splat. DGPR SGU CD ↓ (a) 2.813 (b) 1.273 (c) 1.212 (d) 1.224 (e) 1.201 (f) 1.142 (g) 1.152 (h) 1.125 [PITH_FULL_IMAGE:figures/full_fig_p…
Figure 7
Figure 7. Figure 7: Visualization of MVS feature splatting. Incorporat￾ing MVS feature splatting aids in achieving higher reconstruction accuracy in challenging regions. Using MVS point clouds achieves a more complete re￾construction than SfM. Note that as shown in Tab. 3, due to the inhe…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Multiview Geometric Regularization of Gaussian Splatting for Accurate Radiance Fields

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A multiview geometric regularization scheme that uses MVS depth, RGB, and normal constraints to improve both geometry and rendering quality of Gaussian Splatting radiance fields.

Reference graph

Works this paper leans on

48 extracted references · 26 canonical work pages · cited by 1 Pith paper

  1. [1]

    Large-scale data for multiple-view stereopsis

    Henrik Aanæs, Rasmus Ramsbøl Jensen, George V ogiatzis, Engin Tola, and Anders Bjorholm Dahl. Large-scale data for multiple-view stereopsis. International Journal of Computer Vision, 120:153–168, 2016. 5, 7, 8

    work page 2016

  2. [2]

    Mvsformer: Learning robust image representations via transformers and temperature-based depth for multi-view stereo

    Chenjie Cao, Xinlin Ren, and Yanwei Fu. Mvsformer: Learning robust image representations via transformers and temperature-based depth for multi-view stereo. arXiv preprint arXiv:2208.02541, 2022. 2

    arXiv 2022

  3. [3]

    Rc- mvsnet: Unsupervised multi-view stereo with neural render- ing

    Di Chang, Alja ˇz Bo ˇziˇc, Tong Zhang, Qingsong Yan, Ying- cong Chen, Sabine S ¨usstrunk, and Matthias Nießner. Rc- mvsnet: Unsupervised multi-view stereo with neural render- ing. In European conference on computer vision, pages 665–

    work page

  4. [4]

    Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction

    Danpeng Chen, Hai Li, Weicai Ye, Yifan Wang, Weijian Xie, Shangjin Zhai, Nan Wang, Haomin Liu, Hujun Bao, and Guofeng Zhang. Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction. arXiv preprint arXiv:2406.06521, 2024. 1, 2, 6, 7

    arXiv 2024

  5. [5]

    Neusg: Neural im- plicit surface reconstruction with 3d gaussian splatting guid- ance

    Hanlin Chen, Chen Li, and Gim Hee Lee. Neusg: Neural im- plicit surface reconstruction with 3d gaussian splatting guid- ance. arXiv preprint arXiv:2312.00846, 2023. 1, 2

    arXiv 2023

  6. [6]

    Transmvs- net: Global context-aware multi-view stereo network with transformers

    Yikang Ding, Wentao Yuan, Qingtian Zhu, Haotian Zhang, Xiangyue Liu, Yuanjiang Wang, and Xiao Liu. Transmvs- net: Global context-aware multi-view stereo network with transformers. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 8585– 8594, 2022. 1, 2

    work page 2022

  7. [7]

    Trim 3d gaussian splatting for accurate geometry representation

    Lue Fan, Yuxue Yang, Minxing Li, Hongsheng Li, and Zhaoxiang Zhang. Trim 3d gaussian splatting for accurate geometry representation. arXiv preprint arXiv:2406.07499,

    arXiv

  8. [8]

    Geo-neus: Geometry-consistent neural implicit surfaces learning for multi-view reconstruction

    Qiancheng Fu, Qingshan Xu, Yew Soon Ong, and Wenbing Tao. Geo-neus: Geometry-consistent neural implicit surfaces learning for multi-view reconstruction. Advances in Neural Information Processing Systems, 35:3403–3416, 2022. 2

    2022

Show all 48 references
  1. [9]

    Are we ready for autonomous driving? the kitti vision benchmark suite

    Andreas Geiger, Philip Lenz, and Raquel Urtasun. Are we ready for autonomous driving? the kitti vision benchmark suite. In 2012 IEEE conference on computer vision and pat- tern recognition, pages 3354–3361. IEEE, 2012. 1

    2012

  2. [10]

    Sugar: Surface- aligned gaussian splatting for efficient 3d mesh reconstruc- tion and high-quality mesh rendering

    Antoine Gu ´edon and Vincent Lepetit. Sugar: Surface- aligned gaussian splatting for efficient 3d mesh reconstruc- tion and high-quality mesh rendering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5354–5363, 2024. 2

    2024

  3. [11]

    2d gaussian splatting for geometrically ac- curate radiance fields

    Binbin Huang, Zehao Yu, Anpei Chen, Andreas Geiger, and Shenghua Gao. 2d gaussian splatting for geometrically ac- curate radiance fields. In ACM SIGGRAPH 2024 Conference Papers, pages 1–11, 2024. 1, 2, 3, 4, 5, 6, 7

    2024

  4. [12]

    Neusurf: On-surface priors for neural sur- face reconstruction from sparse input views

    Han Huang, Yulun Wu, Junsheng Zhou, Ge Gao, Ming Gu, and Yu-Shen Liu. Neusurf: On-surface priors for neural sur- face reconstruction from sparse input views. In Proceed- ings of the AAAI Conference on Artificial Intelligence, pages 2312–2320, 2024. 2, 3, 6, 8

    2024

  5. [13]

    Poisson surface reconstruction

    Michael Kazhdan, Matthew Bolitho, and Hugues Hoppe. Poisson surface reconstruction. In Proceedings of the fourth Eurographics symposium on Geometry processing, 2006. 2

    2006

  6. [14]

    3d gaussian splatting for real-time radiance field rendering

    Bernhard Kerbl, Georgios Kopanas, Thomas Leimk ¨uhler, and George Drettakis. 3d gaussian splatting for real-time radiance field rendering. ACM Trans. Graph., 42(4):139–1,

  7. [15]

    Learning unsupervised multi-view stereopsis via robust photometric consistency

    Tejas Khot, Shubham Agrawal, Shubham Tulsiani, Christoph Mertz, Simon Lucey, and Martial Hebert. Learning unsupervised multi-view stereopsis via robust photometric consistency. arXiv preprint arXiv:1905.02706,

    1905 arXiv

  8. [16]

    Learning to fuse monocular and multi-view cues for multi- frame depth estimation in dynamic scenes

    Rui Li, Dong Gong, Wei Yin, Hao Chen, Yu Zhu, Kaix- uan Wang, Xiaozhi Chen, Jinqiu Sun, and Yanning Zhang. Learning to fuse monocular and multi-view cues for multi- frame depth estimation in dynamic scenes. InProceedings of the IEEE/CVF Conference on Computer Vision and Patter...

    2023

  9. [17]

    Know your neighbors: Improving single-view reconstruction via spatial vision- language reasoning

    Rui Li, Tobias Fischer, Mattia Segu, Marc Pollefeys, Luc Van Gool, and Federico Tombari. Know your neighbors: Improving single-view reconstruction via spatial vision- language reasoning. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pa...

    2024

  10. [18]

    Bevdepth: Acquisition of reliable depth for multi-view 3d object detec- tion

    Yinhao Li, Zheng Ge, Guanyi Yu, Jinrong Yang, Zengran Wang, Yukang Shi, Jianjian Sun, and Zeming Li. Bevdepth: Acquisition of reliable depth for multi-view 3d object detec- tion. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 1477–1485, 2023. 1

    2023

  11. [19]

    Retr: Modeling rendering via transformer for generalizable neural surface re- construction

    Yixun Liang, Hao He, and Yingcong Chen. Retr: Modeling rendering via transformer for generalizable neural surface re- construction. Advances in Neural Information Processing Systems, 36, 2024. 1, 3, 6

    2024

  12. [20]

    Sparseneus: Fast generalizable neural sur- face reconstruction from sparse views

    Xiaoxiao Long, Cheng Lin, Peng Wang, Taku Komura, and Wenping Wang. Sparseneus: Fast generalizable neural sur- face reconstruction from sparse views. In European Confer- ence on Computer Vision, pages 210–227. Springer, 2022. 1, 2, 5, 6, 7, 8

    2022

  13. [21]

    Consistent video depth estimation.ACM Transactions on Graphics (ToG), 39(4):71–1, 2020

    Xuan Luo, Jia-Bin Huang, Richard Szeliski, Kevin Matzen, and Johannes Kopf. Consistent video depth estimation.ACM Transactions on Graphics (ToG), 39(4):71–1, 2020. 1

    2020

  14. [22]

    Nerf: Representing scenes as neural radiance fields for view syn- thesis

    Ben Mildenhall, Pratul P Srinivasan, Matthew Tancik, Jonathan T Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. Communications of the ACM, 65(1):99–106, 2021. 1, 2

    2021

  15. [23]

    Unisurf: Unifying neural implicit surfaces and radiance fields for multi-view reconstruction

    Michael Oechsle, Songyou Peng, and Andreas Geiger. Unisurf: Unifying neural implicit surfaces and radiance fields for multi-view reconstruction. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 5589–5599, 2021. 1

    2021

  16. [24]

    Feature splatting: Language-driven physics-based scene syn- thesis and editing

    Ri-Zhao Qiu, Ge Yang, Weijia Zeng, and Xiaolong Wang. Feature splatting: Language-driven physics-based scene syn- thesis and editing. arXiv preprint arXiv:2404.01223, 2024. 4

    2024 arXiv

  17. [25]

    V olrecon: V olume rendering of signed ray distance functions for generalizable multi-view recon- struction

    Yufan Ren, Fangjinhua Wang, Tong Zhang, Marc Pollefeys, and Sabine S¨usstrunk. V olrecon: V olume rendering of signed ray distance functions for generalizable multi-view recon- struction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pa...

    2023

  18. [26]

    Structure- from-motion revisited

    Johannes L Schonberger and Jan-Michael Frahm. Structure- from-motion revisited. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 4104–4113, 2016. 3, 6

    2016

  19. [27]

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

    Cheng Sun, Min Sun, and Hwann-Tzong Chen. Direct voxel grid optimization: Super-fast convergence for radiance fields reconstruction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 5459– 5469, 2022. 2

    2022

  20. [28]

    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, 2021. 1, 2, 6

    2021 arXiv

  21. [29]

    Improved surface reconstruction using high-frequency details

    Yiqun Wang, Ivan Skorokhodov, and Peter Wonka. Improved surface reconstruction using high-frequency details. 2022. 2

    2022

  22. [30]

    Gomvs: Geometrically consistent cost aggregation for multi-view stereo

    Jiang Wu, Rui Li, Haofei Xu, Wenxun Zhao, Yu Zhu, Jinqiu Sun, and Yanning Zhang. Gomvs: Geometrically consistent cost aggregation for multi-view stereo. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20207–20216, 2024. 1, 2

    2024

  23. [31]

    Cl-mvsnet: Unsupervised multi-view stereo with dual-level contrastive learning

    Kaiqiang Xiong, Rui Peng, Zhe Zhang, Tianxing Feng, Jianbo Jiao, Feng Gao, and Ronggang Wang. Cl-mvsnet: Unsupervised multi-view stereo with dual-level contrastive learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 3769–3780, 2023. 1, 2...

    2023

  24. [32]

    Multi-camera collaborative depth prediction via consistent structure esti- mation

    Jialei Xu, Xianming Liu, Yuanchao Bai, Junjun Jiang, Kaix- uan Wang, Xiaozhi Chen, and Xiangyang Ji. Multi-camera collaborative depth prediction via consistent structure esti- mation. In Proceedings of the 30th ACM international con- ference on multimedia, pages 2730–2738, 2022. 2

    2022

  25. [33]

    Sdge: Stereo guided depth estimation for 360° camera sets

    Jialei Xu, Wei Yin, Dong Gong, Junjun Jiang, and Xianming Liu. Sdge: Stereo guided depth estimation for 360° camera sets. In 2024 IEEE/RSJ International Conference on Intelli- gent Robots and Systems (IROS), pages 11179–11186. IEEE,

    2024

  26. [34]

    C2f2neus: Cascade cost frustum fusion for high fidelity and generalizable neu- ral surface reconstruction

    Luoyuan Xu, Tao Guan, Yuesong Wang, Wenkai Liu, Zhao- jie Zeng, Junle Wang, and Wei Yang. C2f2neus: Cascade cost frustum fusion for high fidelity and generalizable neu- ral surface reconstruction. In Proceedings of the IEEE/CVF International Conference on Computer Vision, page...

    2023

  27. [35]

    Learning inverse depth regression for pix- elwise visibility-aware multi-view stereo networks

    Qingshan Xu, Wanjuan Su, Yuhang Qi, Wenbing Tao, and Marc Pollefeys. Learning inverse depth regression for pix- elwise visibility-aware multi-view stereo networks. Interna- tional Journal of Computer Vision, 130(8):2040–2059, 2022. 2

    2022

  28. [36]

    Mvsnet: Depth inference for unstructured multi-view stereo

    Yao Yao, Zixin Luo, Shiwei Li, Tian Fang, and Long Quan. Mvsnet: Depth inference for unstructured multi-view stereo. In Proceedings of the European conference on computer vi- sion (ECCV), pages 767–783, 2018. 1, 2

    2018

  29. [37]

    Blendedmvs: A large- scale dataset for generalized multi-view stereo networks

    Yao Yao, Zixin Luo, Shiwei Li, Jingyang Zhang, Yufan Ren, Lei Zhou, Tian Fang, and Long Quan. Blendedmvs: A large- scale dataset for generalized multi-view stereo networks. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, pages 1790–1799, 2020. 5

    2020

  30. [38]

    V ol- ume rendering of neural implicit surfaces

    Lior Yariv, Jiatao Gu, Yoni Kasten, and Yaron Lipman. V ol- ume rendering of neural implicit surfaces. Advances in Neu- ral Information Processing Systems, 34:4805–4815, 2021. 1, 2, 6

    2021

  31. [39]

    Fast normalized cross- correlation

    Jae-Chern Yoo and Tae Hee Han. Fast normalized cross- correlation. Circuits, systems and signal processing, 28:819– 843, 2009. 5

    2009

  32. [40]

    Spar- secraft: Few-shot neural reconstruction through stereopsis guided geometric linearization

    Mae Younes, Amine Ouasfi, and Adnane Boukhayma. Spar- secraft: Few-shot neural reconstruction through stereopsis guided geometric linearization. In European Conference on Computer Vision, pages 37–56. Springer, 2024. 3, 6

    2024

  33. [41]

    Gsdf: 3dgs meets sdf for improved rendering and reconstruction

    Mulin Yu, Tao Lu, Linning Xu, Lihan Jiang, Yuanbo Xiangli, and Bo Dai. Gsdf: 3dgs meets sdf for improved rendering and reconstruction. arXiv preprint arXiv:2403.16964, 2024. 1, 2

    2024 arXiv

  34. [42]

    Monosdf: Exploring monocu- lar geometric cues for neural implicit surface reconstruc- tion

    Zehao Yu, Songyou Peng, Michael Niemeyer, Torsten Sat- tler, and Andreas Geiger. Monosdf: Exploring monocu- lar geometric cues for neural implicit surface reconstruc- tion. Advances in neural information processing systems , 35:25018–25032, 2022. 2

    2022

  35. [43]

    Gaussian opacity fields: Efficient and compact surface reconstruc- tion in unbounded scenes

    Zehao Yu, Torsten Sattler, and Andreas Geiger. Gaussian opacity fields: Efficient and compact surface reconstruc- tion in unbounded scenes. arXiv preprint arXiv:2404.10772,

    arXiv

  36. [44]

    Rade-gs: Rasterizing depth in gaussian splatting

    Baowen Zhang, Chuan Fang, Rakesh Shrestha, Yixun Liang, Xiaoxiao Long, and Ping Tan. Rade-gs: Rasterizing depth in gaussian splatting. arXiv preprint arXiv:2406.01467, 2024. 2

    2024 arXiv

  37. [45]

    Learning signed distance field for multi-view surface reconstruction

    Jingyang Zhang, Yao Yao, and Long Quan. Learning signed distance field for multi-view surface reconstruction. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 6525–6534, 2021. 5

    2021

  38. [46]

    Vis-mvsnet: Visibility-aware multi-view stereo net- work

    Jingyang Zhang, Shiwei Li, Zixin Luo, Tian Fang, and Yao Yao. Vis-mvsnet: Visibility-aware multi-view stereo net- work. International Journal of Computer Vision , 131(1): 199–214, 2023. 2

    2023

  39. [47]

    Feature 3dgs: Supercharging 3d gaussian splatting to enable distilled feature fields

    Shijie Zhou, Haoran Chang, Sicheng Jiang, Zhiwen Fan, Ze- hao Zhu, Dejia Xu, Pradyumna Chari, Suya You, Zhangyang Wang, and Achuta Kadambi. Feature 3dgs: Supercharging 3d gaussian splatting to enable distilled feature fields. InPro- ceedings of the IEEE/CVF Conference on Compu...

    2024

  40. [680]

    Springer, 2022. 2, 5

    2022

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.