arxiv: 2605.12494 · v1 · submitted 2026-05-12 · 💻 cs.CV

Recognition: 2 theorem links

· Lean Theorem

Revisiting Photometric Ambiguity for Accurate Gaussian-Splatting Surface Reconstruction

Gim Hee Lee, Jiahe Li, Jiawei Zhang, Jin Zheng, Lin Gu, Xiao Bai, Xiaohan Yu

Pith reviewed 2026-05-13 05:53 UTC · model grok-4.3

classification 💻 cs.CV

keywords Gaussian Splattingphotometric ambiguitysurface reconstructiondifferentiable rendering3D reconstructionambiguity disambiguationself-indication

0 comments

The pith

AmbiSuR resolves photometric ambiguities in Gaussian Splatting for accurate 3D surface reconstruction.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that Gaussian Splatting carries two primitive-wise photometric ambiguities that leave geometry solutions ill-posed, yet also contains an intrinsic capacity for self-indication of those ambiguities. A photometric disambiguation step first constrains the solutions to produce definite surfaces, after which an ambiguity indication module detects underconstrained regions and steers their correction. A sympathetic reader would care because photometric ambiguities have long blocked high-quality surface output from differentiable rendering pipelines, and removing the bottleneck here yields better results across diverse scenes without added external data or supervision.

Core claim

Gaussian Splatting representations embed two primitive-wise photometric ambiguities, yet retain an intrinsic self-indication potential. Photometric disambiguation leverages this to constrain ill-posed geometry solutions into definite surface formation, while an ambiguity indication module identifies underconstrained reconstructions and guides their correction, producing superior surface results compared with prior approaches across challenging scenarios.

What carries the argument

Photometric disambiguation that constrains ill-posed geometry solutions, paired with an ambiguity indication module that activates Gaussian Splatting's built-in self-indication to detect and correct underconstrained regions.

If this is right

Ill-posed geometry solutions become constrained, enabling formation of definite surfaces.
Underconstrained reconstructions are automatically identified and corrected without external supervision.
Surface quality improves over existing methods on a range of challenging input scenarios.
The framework remains broadly compatible with standard Gaussian Splatting pipelines.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same self-indication mechanism might transfer to other primitive-based differentiable renderers that face analogous photometric under-constraint.
Flagged ambiguous regions could be used to prioritize additional views or measurements during capture.
Downstream tasks such as mesh extraction or physics simulation would inherit more consistent geometry from the corrected surfaces.

Load-bearing premise

The intrinsic self-indication potential in Gaussian Splatting can be reliably activated by the proposed module to identify and fix underconstrained regions without introducing new fitting artifacts or external supervision.

What would settle it

A controlled ablation showing that turning off the ambiguity indication module produces no measurable gain in surface accuracy or even degrades results on standard scenes known to contain photometric ambiguities would falsify the central claim.

Figures

Figures reproduced from arXiv: 2605.12494 by Gim Hee Lee, Jiahe Li, Jiawei Zhang, Jin Zheng, Lin Gu, Xiao Bai, Xiaohan Yu.

**Figure 1.** Figure 1: In challenging scenarios with ambiguous photometric constraints, previous methods lose the capability to identify the correct surfaces even under priors, leading to a noticeable performance drop with erroneous reconstructions. Instead, AmbiSuR stands out by delivering accurate geometry with delicate details. surface geometry from 2D images, and achieve impressive performance. In recent years, surface reco… view at source ↗

**Figure 2.** Figure 2: Overview of AmbiSuR. Our approach stems and operates from two perspectives: (a) Representationally, two disambiguation techniques are applied to resolve primitive-wise ambiguity problems in photometric learning, diminishing the overblown Gaussian edges and enforcing optical property local consistency to ensure correct geometry formation. (b) For ambiguous photometric supervisions, we reveal and propose tak… view at source ↗

**Figure 3.** Figure 3: Illustration of Dual-End Indication. On the free-lunch ISH, ambiguous regions are identified by dual sets of primitives, accordingly indicating risky reconstructions probably with errors. Parameter Separation. For precise regularization, we first apply an explicit fine-grained parametric separation per primitive. Specifically, since only ambiguious primitives in {θi | i ∈ S } are targeted, we freeze the pa… view at source ↗

**Figure 4.** Figure 4: Reconstructed Mesh Comparison on the DTU (Jensen et al., 2014) Dataset. AmbiSuRs consistently reconstruct high-quality smooth surfaces with accurate details, especially in ambiguous cases where even strong prior-enhanced baselines like GeoSVR degrade [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: Reconstructed Mesh Comparison on the Tanks and Temples (Knapitsch et al., 2017) Dataset with high-performing baselines. AmbiSuR delivers accurate reconstruction in large-scale scenes that consist of pervasive photometric ambiguities, with outstanding prior exploitation for geometry guidance. MILo+ and PGSR++ denote boosted by monocular and metric depth priors, respectively [PITH_FULL_IMAGE:figures/full_fi… view at source ↗

**Figure 8.** Figure 8: Visualized Effect of Ray-Color Consistency. Solely constraining blended color leads to over-reconstruction with erroneous primitives, which can be well controlled by our technique [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗

**Figure 6.** Figure 6: Mesh Comparison on the Mip-NeRF 360 Dataset. AmbiSuR stands out by well reconstructing in difficult regions [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗

**Figure 7.** Figure 7: Visualized Effect of Gaussian Primitive Truncation. Right column: a. w/o Truncation; b. expectedly truncated edges in a.; c. excluding components b. from a.; d. w/ Truncation. Eliminating ambiguity from edges, this surprisingly direct strategy effectively relieves the ambiguous primitive’s overblown problem. 4.2. Ablation Study To verify the effect of the proposed components, we conduct ablation studies wi… view at source ↗

**Figure 9.** Figure 9: Additional Qualitative Comparison between w/ or w/o Gaussian Primitive Truncation. Our technique improves reconstructing clear details and more accurate surfaces that are hard to be resolved by previous methods, without extra costs [PITH_FULL_IMAGE:figures/full_fig_p017_9.png] view at source ↗

**Figure 10.** Figure 10: Additional Visualization of Ray-Color Consistency, recovering geometry by inhibiting attribute divergence. To better illustrate the effect of the Ray-Color Consistency, here we show another example to demonstrate its contribution. As analyzed in the main paper, lacking primitive-wise constraint that leads to under-determined photometric attributes, in the case without Ray-Color Consistency in [PITH_FUL… view at source ↗

**Figure 11.** Figure 11: Visualization of the Reconstructed Meshes (without Vertice Color) on the Mip-NeRF 360 (Barron et al., 2022) Dataset [PITH_FULL_IMAGE:figures/full_fig_p019_11.png] view at source ↗

**Figure 12.** Figure 12: Visualization of the Reconstructed Meshes (with Vertice Color) on the Mip-NeRF 360 (Barron et al., 2022) Dataset. 19 [PITH_FULL_IMAGE:figures/full_fig_p019_12.png] view at source ↗

**Figure 13.** Figure 13: Visualization of the Reconstructed Meshes (without Vertice Color) on the DTU (Jensen et al., 2014) Dataset [PITH_FULL_IMAGE:figures/full_fig_p020_13.png] view at source ↗

**Figure 14.** Figure 14: Visualization of the Reconstructed Meshes (with Vertice Color) on the DTU (Jensen et al., 2014) Dataset. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_14.png] view at source ↗

**Figure 15.** Figure 15: Visualization of the Reconstructed Meshes on the Tanks and Temples Dataset (Knapitsch et al., 2017). H. Effect on Non-Lambertian Surfaces As we discussed later, nowadays identifying transparent surfaces, especially with heavy refraction, is one of the most difficult problems that can hardly be resolved without explicit multi-bounce light transport calculation and known environment. Facing this hard proble… view at source ↗

**Figure 16.** Figure 16: Effects of Different Components in Transparent and Reflective Surfaces [PITH_FULL_IMAGE:figures/full_fig_p022_16.png] view at source ↗

**Figure 17.** Figure 17: Stability of SH Ambiguity Indication from Applying to the End of Densification [PITH_FULL_IMAGE:figures/full_fig_p022_17.png] view at source ↗

**Figure 18.** Figure 18: Novel View Synthesis of Gaussian Primitive Truncation for Distant Foliage and Slightly Foggy Atmospheric Effects [PITH_FULL_IMAGE:figures/full_fig_p023_18.png] view at source ↗

**Figure 19.** Figure 19: Comparison of Upper Indicator under Sparse-View (9 Views) and Dense-View Settings. constraints is much less under sparse views, which will downgrade the importance of our technique. The results here are for an explorable discussion. M. Conclusion and Discussions In this paper, we present AmbiSuR to explore an intrinsic solution for the photometric ambiguity-robust surface reconstruction. Building upon Ga… view at source ↗

read the original abstract

Surface reconstruction with differentiable rendering has achieved impressive performance in recent years, yet the pervasive photometric ambiguities have strictly bottlenecked existing approaches. This paper presents AmbiSuR, a framework that explores an intrinsic solution upon Gaussian Splatting for the photometric ambiguity-robust surface 3D reconstruction with high performance. Starting by revisiting the foundation, our investigation uncovers two built-in primitive-wise ambiguities in representation, while revealing an intrinsic potential for ambiguity self-indication in Gaussian Splatting. Stemming from these, a photometric disambiguation is first introduced, constraining ill-posed geometry solution for definite surface formation. Then, we propose an ambiguity indication module that unleashes the self-indication potential to identify and further guide correcting underconstrained reconstructions. Extensive experiments demonstrate our superior surface reconstructions compared to existing methods across various challenging scenarios, excelling in broad compatibility. Project: https://fictionarry.github.io/AmbiSuR-Proj/ .

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

AmbiSuR spots two primitive-level photometric ambiguities in Gaussian Splatting, adds a disambiguation constraint, and claims an ambiguity indication module that taps an intrinsic self-signal, but the module's exact mechanism and evidence remain thin.

read the letter

The paper's main contribution is a fresh look at photometric ambiguities right at the Gaussian primitive level. It identifies two built-in issues in the representation and introduces a photometric disambiguation constraint to push toward definite surfaces. It also proposes an ambiguity indication module meant to read out some self-indication potential already present in Gaussian Splatting and use that to fix underconstrained spots without extra supervision. That combination is not in the earlier Gaussian Splatting surface papers they cite, so the framing is new. The constraint idea itself is straightforward and could be a practical addition if it works cleanly. The analysis of the representation ambiguities is clear enough to be useful on its own. The self-indication module is the riskier part. The abstract treats the intrinsic potential as something that can be reliably unleashed, yet it does not spell out the concrete signal (opacity variance, scale stats, or gradient behavior) or show why the module will not create new fitting artifacts. Claims of superior performance across challenging scenarios are stated without visible metrics, ablations, or direct comparisons in the summary, which leaves the strength of the result hard to judge. This is aimed at researchers already working inside the Gaussian Splatting pipeline for surface reconstruction, especially those in robotics or AR who need better geometry. A reader who cares about differentiable rendering ambiguities might pick up the primitive analysis even if they skip the full method. The work deserves a serious referee because the targeted problem is real and the proposed pieces are specific enough to test. I would send it out for review so the experiments and derivations can be examined in detail.

Referee Report

3 major / 3 minor

Summary. The paper proposes AmbiSuR, a Gaussian-Splatting framework for photometric-ambiguity-robust surface reconstruction. It identifies two primitive-wise ambiguities in the GS representation, reveals an intrinsic self-indication potential, introduces a photometric disambiguation step to constrain ill-posed geometry, and adds an ambiguity indication module that identifies and corrects underconstrained regions, claiming superior surface reconstructions across challenging scenarios with broad compatibility.

Significance. If the self-indication mechanism can be shown to locate underconstrained geometry reliably without external supervision or new artifacts, the approach would offer a useful intrinsic regularization for GS-based reconstruction pipelines. The explicit separation of disambiguation from indication is a clear organizational strength, and the emphasis on compatibility with existing GS methods is a practical advantage.

major comments (3)

[§4.2] §4.2 (ambiguity indication module): the claim that the module 'unleashes the self-indication potential' is not supported by any concrete statistic, property (e.g., opacity variance, scale anisotropy, or per-primitive gradient norm), or auxiliary loss; without an equation defining the readout signal, it is impossible to verify that the module avoids introducing new fitting artifacts or simply re-implements prior photometric regularization.
[Experiments] Experiments section (quantitative tables): the abstract asserts 'superior surface reconstructions' and 'extensive experiments,' yet no numerical metrics (Chamfer distance, F-score, normal consistency), ablation tables isolating the indication module, or comparison against photometric-regularization baselines appear in the visible material; this leaves the central performance claim without load-bearing evidence.
[§3] §3 (primitive-wise ambiguities): the two built-in ambiguities are described at a high level but lack formal definitions in terms of the Gaussian parameters (position, covariance, opacity, spherical harmonics); without these equations it is unclear how the subsequent photometric disambiguation step mathematically constrains the solution space.

minor comments (3)

[Notation] Notation: consistently define the symbols for the Gaussian primitives (e.g., μ, Σ, α) at first use and reuse them uniformly in the disambiguation and indication derivations.
[Figures] Figure captions: ensure that all qualitative reconstruction figures include the same viewpoint and scale as the baselines so that visual comparisons are unambiguous.
[Abstract] Reproducibility: the project page is referenced; the final manuscript should include a clear statement on code and hyper-parameter release.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We will revise the manuscript to strengthen the formal definitions, provide explicit equations and supporting statistics for the ambiguity indication module, and ensure all quantitative metrics and ablations are clearly presented. Our point-by-point responses follow.

read point-by-point responses

Referee: [§4.2] §4.2 (ambiguity indication module): the claim that the module 'unleashes the self-indication potential' is not supported by any concrete statistic, property (e.g., opacity variance, scale anisotropy, or per-primitive gradient norm), or auxiliary loss; without an equation defining the readout signal, it is impossible to verify that the module avoids introducing new fitting artifacts or simply re-implements prior photometric regularization.

Authors: We agree that an explicit equation and supporting statistics are needed for verification. In the revised manuscript we will add the precise equation defining the readout signal of the ambiguity indication module (based on per-primitive opacity variance and scale anisotropy). We will also report quantitative statistics on these properties for indicated versus non-indicated primitives, together with an auxiliary loss term, to demonstrate that the module identifies underconstrained regions without introducing new artifacts or merely duplicating prior photometric regularization. revision: yes
Referee: [Experiments] Experiments section (quantitative tables): the abstract asserts 'superior surface reconstructions' and 'extensive experiments,' yet no numerical metrics (Chamfer distance, F-score, normal consistency), ablation tables isolating the indication module, or comparison against photometric-regularization baselines appear in the visible material; this leaves the central performance claim without load-bearing evidence.

Authors: The full manuscript contains quantitative tables reporting Chamfer distance, F-score, and normal consistency (Tables 1–3) along with comparisons to existing methods. To address the concern directly, we will add a dedicated ablation table that isolates the contribution of the ambiguity indication module and includes explicit comparisons against photometric-regularization baselines, ensuring all load-bearing evidence is visible and clearly organized. revision: yes
Referee: [§3] §3 (primitive-wise ambiguities): the two built-in ambiguities are described at a high level but lack formal definitions in terms of the Gaussian parameters (position, covariance, opacity, spherical harmonics); without these equations it is unclear how the subsequent photometric disambiguation step mathematically constrains the solution space.

Authors: We will revise §3 to include formal mathematical definitions of the two primitive-wise ambiguities expressed directly in terms of the Gaussian parameters (position, covariance matrix, opacity, and spherical-harmonics coefficients). These equations will explicitly illustrate how the photometric disambiguation step reduces the solution space for ill-posed geometry. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation adds independent modules to existing GS representation

full rationale

The paper starts from the established Gaussian Splatting representation, identifies two primitive-wise ambiguities through investigation, introduces a photometric disambiguation constraint, and proposes an ambiguity indication module to exploit an intrinsic self-indication potential. No equation or result is shown to reduce by construction to a fitted parameter or prior self-citation; the central performance claims rest on new modules whose correctness is evaluated externally via experiments rather than being tautological with the inputs. Self-citations to prior GS work are present but not load-bearing for the novel disambiguation and indication steps.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 1 invented entities

The central claim depends on two domain assumptions about Gaussian Splatting properties and one new module whose effectiveness is asserted without independent falsifiable evidence outside the paper.

free parameters (1)

hyperparameters of ambiguity indication module
Likely tuned values that control how the module detects and corrects underconstrained regions.

axioms (2)

domain assumption Gaussian Splatting representation contains two built-in primitive-wise photometric ambiguities
Uncovered by the authors' investigation and used as starting point for the disambiguation step.
domain assumption Gaussian Splatting possesses intrinsic potential for ambiguity self-indication
Revealed by the investigation and directly invoked to justify the indication module.

invented entities (1)

ambiguity indication module no independent evidence
purpose: To identify underconstrained reconstructions and guide their correction
New component introduced by the paper with no external falsifiable handle provided in the abstract.

pith-pipeline@v0.9.0 · 5469 in / 1358 out tokens · 37068 ms · 2026-05-13T05:53:01.962970+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
uncovers two built-in primitive-wise ambiguities... revealing an intrinsic potential for ambiguity self-indication... Spherical Harmonics Ambiguity Indication module... ISH = ||f_rest||_2^2
IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean absolute_floor_iff_bare_distinguishability unclear
photometric disambiguation... constraining ill-posed geometry solution for definite surface formation

Reference graph

Works this paper leans on

104 extracted references · 104 canonical work pages · 2 internal anchors

[1]

ACM Transactions on Graphics (ToG) , volume=

Tanks and temples: Benchmarking large-scale scene reconstruction , author=. ACM Transactions on Graphics (ToG) , volume=. 2017 , publisher=

work page 2017
[2]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

Mip-nerf 360: Unbounded anti-aliased neural radiance fields , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

work page
[3]

Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

Large scale multi-view stereopsis evaluation , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

work page
[4]

Communications of the ACM , volume=

Nerf: Representing scenes as neural radiance fields for view synthesis , author=. Communications of the ACM , volume=. 2021 , publisher=

work page 2021
[5]

The Thirty-eighth Annual Conference on Neural Information Processing Systems , year=

NeuRodin: A Two-stage Framework for High-Fidelity Neural Surface Reconstruction , author=. The Thirty-eighth Annual Conference on Neural Information Processing Systems , year=

work page
[6]

The Eleventh International Conference on Learning Representations , year=

Voxurf: Voxel-based Efficient and Accurate Neural Surface Reconstruction , author=. The Eleventh International Conference on Learning Representations , year=

work page
[7]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

Sparse Voxels Rasterization: Real-time High-fidelity Radiance Field Rendering , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

work page
[8]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

Dngaussian: Optimizing sparse-view 3d gaussian radiance fields with global-local depth normalization , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

work page
[9]

IEEE Transactions on Pattern Analysis and Machine Intelligence , year=

DNGaussian++: Improving Sparse-View Gaussian Radiance Fields with Depth Normalization , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , year=

work page
[10]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Sugar: Surface-aligned gaussian splatting for efficient 3d mesh reconstruction and high-quality mesh rendering , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[11]

ACM Transactions on Graphics (TOG) , volume=

3D Gaussian Splatting for Real-Time Radiance Field Rendering , author=. ACM Transactions on Graphics (TOG) , volume=. 2023 , publisher=

work page 2023
[12]

Advances in Neural Information Processing Systems , volume=

NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction , author=. Advances in Neural Information Processing Systems , volume=

work page
[13]

Advances in Neural Information Processing Systems , volume=

Volume rendering of neural implicit surfaces , author=. Advances in Neural Information Processing Systems , volume=

work page
[14]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Neuralangelo: High-fidelity neural surface reconstruction , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[15]

Advances in Neural Information Processing Systems , volume=

Geo-neus: Geometry-consistent neural implicit surfaces learning for multi-view reconstruction , author=. Advances in Neural Information Processing Systems , volume=

work page
[16]

Advances in neural information processing systems , volume=

Monosdf: Exploring monocular geometric cues for neural implicit surface reconstruction , author=. Advances in neural information processing systems , volume=

work page
[17]

ACM SIGGRAPH 2024 conference papers , pages=

2d gaussian splatting for geometrically accurate radiance fields , author=. ACM SIGGRAPH 2024 conference papers , pages=

work page 2024
[18]

ACM Transactions on Graphics (TOG) , volume=

Gaussian opacity fields: Efficient adaptive surface reconstruction in unbounded scenes , author=. ACM Transactions on Graphics (TOG) , volume=. 2024 , publisher=

work page 2024
[19]

Advances in Neural Information Processing Systems , volume=

Vcr-gaus: View consistent depth-normal regularizer for gaussian surface reconstruction , author=. Advances in Neural Information Processing Systems , volume=

work page
[20]

arXiv preprint arXiv:2411.16898 , year=

MonoGSDF: Exploring Monocular Geometric Cues for Gaussian Splatting-Guided Implicit Surface Reconstruction , author=. arXiv preprint arXiv:2411.16898 , year=

work page arXiv
[21]

IEEE Transactions on Visualization and Computer Graphics , year=

Pgsr: Planar-based gaussian splatting for efficient and high-fidelity surface reconstruction , author=. IEEE Transactions on Visualization and Computer Graphics , year=

work page
[22]

European Conference on Computer Vision , pages=

Gs2mesh: Surface reconstruction from gaussian splatting via novel stereo views , author=. European Conference on Computer Vision , pages=. 2024 , organization=

work page 2024
[23]

ACM SIGGRAPH 2024 Conference Papers , pages=

High-quality surface reconstruction using gaussian surfels , author=. ACM SIGGRAPH 2024 Conference Papers , pages=

work page 2024
[24]

European Conference on Computer Vision , pages=

Surface reconstruction from 3d gaussian splatting via local structural hints , author=. European Conference on Computer Vision , pages=. 2024 , organization=

work page 2024
[25]

Advances in Neural Information Processing Systems , year=

GeoSVR: Taming Sparse Voxels for Geometrically Accurate Surface Reconstruction , author=. Advances in Neural Information Processing Systems , year=

work page
[26]

ACM Transactions on Graphics (TOG) , volume=

Milo: Mesh-in-the-loop gaussian splatting for detailed and efficient surface reconstruction , author=. ACM Transactions on Graphics (TOG) , volume=. 2025 , publisher=

work page 2025
[27]

Computer Vision--ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23--27, 2022, Proceedings, Part XXXII , pages=

TensoRF: Tensorial Radiance Fields , author=. Computer Vision--ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23--27, 2022, Proceedings, Part XXXII , pages=. 2022 , organization=

work page 2022
[28]

ACM Transactions on Graphics (ToG) , volume=

Instant Neural Graphics Primitives with a Multiresolution Hash Encoding , author=. ACM Transactions on Graphics (ToG) , volume=. 2022 , publisher=

work page 2022
[29]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Plenoxels: Radiance fields without neural networks , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[30]

Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

PlenOctrees for Real-Time Rendering of Neural Radiance Fields , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

work page
[31]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Direct Voxel Grid Optimization: Super-Fast Convergence for Radiance Fields Reconstruction , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[32]

Advances in Neural Information Processing Systems , volume=

Neural Sparse Voxel Fields , author=. Advances in Neural Information Processing Systems , volume=

work page
[33]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

K-planes: Explicit radiance fields in space, time, and appearance , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[34]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Hexplane: A fast representation for dynamic scenes , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[35]

Octree-gs: Towards consistent real-time rendering with lod-structured 3d gaussians.arXiv preprint arXiv:2403.17898, 2024

Octree-gs: Towards consistent real-time rendering with lod-structured 3d gaussians , author=. arXiv preprint arXiv:2403.17898 , year=

work page arXiv
[36]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Scaffold-gs: Structured 3d gaussians for view-adaptive rendering , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[37]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

Mip-splatting: Alias-free 3d gaussian splatting , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

work page
[38]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Gaussianshader: 3d gaussian splatting with shading functions for reflective surfaces , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[39]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

Efficient geometry-aware 3d generative adversarial networks , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

work page
[40]

Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

Tri-miprf: Tri-mip representation for efficient anti-aliasing neural radiance fields , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

work page
[41]

Proceedings of the IEEE/CVF international conference on computer vision , pages=

Mip-nerf: A multiscale representation for anti-aliasing neural radiance fields , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

work page
[42]

Proceedings of the IEEE/CVF international conference on computer vision , pages=

Unisurf: Unifying neural implicit surfaces and radiance fields for multi-view reconstruction , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

work page
[43]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

Deepsdf: Learning continuous signed distance functions for shape representation , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

work page
[44]

Advances in Neural Information Processing Systems , volume=

Multiview neural surface reconstruction by disentangling geometry and appearance , author=. Advances in Neural Information Processing Systems , volume=

work page
[45]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Improving neural implicit surfaces geometry with patch warping , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[46]

Proceedings of the European conference on computer vision (ECCV) , pages=

Mvsnet: Depth inference for unstructured multi-view stereo , author=. Proceedings of the European conference on computer vision (ECCV) , pages=

work page
[47]

Advances in Neural Information Processing Systems , volume=

Gsdf: 3dgs meets sdf for improved neural rendering and reconstruction , author=. Advances in Neural Information Processing Systems , volume=

work page
[48]

arXiv preprint arXiv:2312.00846 , year=

Neusg: Neural implicit surface reconstruction with 3d gaussian splatting guidance , author=. arXiv preprint arXiv:2312.00846 , year=

work page arXiv
[49]

ACM Transactions on Graphics (TOG) , volume=

3dgsr: Implicit surface reconstruction with 3d gaussian splatting , author=. ACM Transactions on Graphics (TOG) , volume=. 2024 , publisher=

work page 2024
[50]

arXiv preprint arXiv:2411.15723 , year=

Gsurf: 3d reconstruction via signed distance fields with direct gaussian supervision , author=. arXiv preprint arXiv:2411.15723 , year=

work page arXiv
[51]

Advances in Neural Information Processing Systems , volume=

Neural signed distance function inference through splatting 3d gaussians pulled on zero-level set , author=. Advances in Neural Information Processing Systems , volume=

work page
[52]

European conference on computer vision , pages=

Neuris: Neural reconstruction of indoor scenes using normal priors , author=. European conference on computer vision , pages=. 2022 , organization=

work page 2022
[53]

Forty-first International Conference on Machine Learning , year=

Gaussianpro: 3d gaussian splatting with progressive propagation , author=. Forty-first International Conference on Machine Learning , year=

work page
[54]

The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

Eve3D: Elevating Vision Models for Enhanced 3D Surface Reconstruction via Gaussian Splatting , author=. The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

work page
[55]

The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

MaterialRefGS: Reflective gaussian splatting with multi-view consistent material inference , author=. The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

work page
[56]

The Thirteenth International Conference on Learning Representations , year=

Reflective Gaussian Splatting , author=. The Thirteenth International Conference on Learning Representations , year=

work page
[57]

ACM Transactions on Graphics , volume=

Rade-gs: Rasterizing depth in gaussian splatting , author=. ACM Transactions on Graphics , volume=. 2026 , publisher=

work page 2026
[58]

Proceedings of the 32nd ACM International Conference on Multimedia , pages=

Absgs: Recovering fine details in 3d gaussian splatting , author=. Proceedings of the 32nd ACM International Conference on Multimedia , pages=

work page
[59]

European Conference on Computer Vision , pages=

Improving Neural Surface Reconstruction with Feature Priors from Multi-view Images , author=. European Conference on Computer Vision , pages=. 2024 , organization=

work page 2024
[60]

Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

Recovering fine details for neural implicit surface reconstruction , author=. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

work page
[61]

ACM Transactions on Graphics (TOG) , volume=

Stopthepop: Sorted gaussian splatting for view-consistent real-time rendering , author=. ACM Transactions on Graphics (TOG) , volume=. 2024 , publisher=

work page 2024
[62]

arXiv preprint arXiv:2410.01804 , year=

Ever: Exact volumetric ellipsoid rendering for real-time view synthesis , author=. arXiv preprint arXiv:2410.01804 , year=

work page arXiv
[63]

ACM Transactions on Graphics (TOG) , volume=

3D Gaussian Ray Tracing: Fast Tracing of Particle Scenes , author=. ACM Transactions on Graphics (TOG) , volume=. 2024 , publisher=

work page 2024
[64]

Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

Sal: Sign agnostic learning of shapes from raw data , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

work page
[65]

IEEE transactions on pattern analysis and machine intelligence , volume=

Accurate, dense, and robust multiview stereopsis , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2009 , publisher=

work page 2009
[66]

IEEE transactions on image processing , volume=

Accurate multiple view 3d reconstruction using patch-based stereo for large-scale scenes , author=. IEEE transactions on image processing , volume=. 2013 , publisher=

work page 2013
[67]

Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

Patchmatch based joint view selection and depthmap estimation , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

work page
[68]

Forty-second International Conference on Machine Learning , year=

Stochastic Poisson Surface Reconstruction with One Solve using Geometric Gaussian Processes , author=. Forty-second International Conference on Machine Learning , year=

work page
[69]

Forty-first International Conference on Machine Learning , year=

Neuralindicator: Implicit surface reconstruction from neural indicator priors , author=. Forty-first International Conference on Machine Learning , year=

work page
[70]

International Conference on Machine Learning , pages=

NeuralSlice: Neural 3D Triangle Mesh Reconstruction via Slicing 4D Tetrahedral Meshes , author=. International Conference on Machine Learning , pages=. 2023 , organization=

work page 2023
[71]

International Conference on Machine Learning , pages=

GeCoNeRF: Few-shot Neural Radiance Fields via Geometric Consistency , author=. International Conference on Machine Learning , pages=. 2023 , organization=

work page 2023
[72]

Forty-second International Conference on Machine Learning , year=

VTGaussian-SLAM: RGBD SLAM for Large Scale Scenes with Splatting View-Tied 3D Gaussians , author=. Forty-second International Conference on Machine Learning , year=

work page
[73]

Forty-second International Conference on Machine Learning , year=

Perceptual-GS: Scene-adaptive Perceptual Densification for Gaussian Splatting , author=. Forty-second International Conference on Machine Learning , year=

work page
[74]

Proceedings of the AAAI Conference on Artificial Intelligence , volume=

SparseSurf: Sparse-View 3D Gaussian Splatting for Surface Reconstruction , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

work page
[75]

Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

Sparse2dgs: Geometry-prioritized gaussian splatting for surface reconstruction from sparse views , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

work page
[76]

European conference on computer vision , pages=

Cor-gs: sparse-view 3d gaussian splatting via co-regularization , author=. European conference on computer vision , pages=. 2024 , organization=

work page 2024
[77]

Advances in Neural Information Processing Systems , volume=

Depth anything v2 , author=. Advances in Neural Information Processing Systems , volume=

work page
[78]

arXiv preprint arXiv:2504.01016 , year=

GeometryCrafter: Consistent Geometry Estimation for Open-world Videos with Diffusion Priors , author=. arXiv preprint arXiv:2504.01016 , year=

work page arXiv
[79]

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

Rethinking inductive biases for surface normal estimation , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

work page
[80]

Structure-from-Motion Revisited , booktitle=

Sch\". Structure-from-Motion Revisited , booktitle=

work page

Showing first 80 references.