Neural Shortest Path for Surface Reconstruction from Point Clouds

Imseong Park; Jooyoung Hahn; Myungjoo Kang; Yesom Park

REVIEW 4 major objections 5 minor 59 references

Neural Shortest Path for Surface Reconstruction from Point Clouds

T0 review · 4 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read A neural field that learns the exact shortest path to a surface, not just its distance, can reconstruct open shapes from noisy point clouds.

desk verdict The F = d∇d representation is a genuine advance for open-surface INR, but the ESP guarantee is proved for a loss that is not the one being trained. read the letter →

arxiv 2502.06047 v1 pith:UBZZIY5G submitted 2025-02-09 cs.LG

Yesom Park , Imseong Park , Jooyoung Hahn , Myungjoo Kang This is my paper · ORCID

classification cs.LG MSC 65D1768T0735F21

keywords surfacereconstructionpointcloudsimplicitneuralrepresentationunsigneddistancefunctionexactshortestpathvariablesplittingeikonalequationH1convergence

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper is trying to establish that a neural network can learn, directly from a point cloud, the exact shortest path to the unknown surface: the vector field that points each query toward its nearest surface point and whose length is the distance to the surface. Its central object is the identity $F(x)=d(x)\nabla d(x)$, so learning $F$ means learning the distance and its gradient at once. The authors prove that when this vector field is decomposed into magnitude and unit direction and trained with their three-part loss, the magnitude converges in the $H^1(\Omega)$ norm, meaning the gradient converges too, and the global minimizer of the idealized loss is exactly the desired shortest-path field. If the proof carries over to the trained model, surface reconstruction gains a principled way to handle open and non-watertight shapes with less sensitivity to noisy or sparse scans, because the method learns geometric paths rather than trusting every input point.

What carries the argument

The load-bearing structure is the magnitude-direction decomposition of the network output, $F_\theta=\|F_\theta\|\,(F_\theta/\|F_\theta\|)=d_\theta G_\theta$, combined with a variable-splitting constraint that makes the unit direction the gradient of the magnitude. Three loss terms drive the mechanism: the manifold loss anchors the magnitude to zero on the given points, the gradient-matching loss enforces $\nabla d_\theta=G_\theta$, and the shortest-path loss requires the pulled point $x-F_\theta(x)$ to lie on the zero level set. The shortest-path term is what forces the learned field to reproduce the identity $F=d\nabla d$, which is the key to the global-minimum theorem, while the decomposition is what makes the $H^1$-convergence theorem for the magnitude possible.

What would settle it

Train the full network with the practical loss on a clean hemisphere point cloud, then compare the learned field against the ground-truth distance $d$ and shortest path $d\nabla d$ on a fine grid away from the surface. If, at convergence, the $H^1(\Omega)$ error $\|d_\theta-d\|_{H^1(\Omega)}$ or the shortest-path error $\|F_\theta-d\nabla d\|_{L^2(\Omega\setminus\Gamma)}$ is bounded away from zero even as the loss approaches zero, then the reported guarantee does not transfer to the deployed training objective.

Watch

Extended reading notes

Core claim

On its own terms, the central discovery is that approximating the exact shortest path $F(x)=d(x)\nabla d(x)$ yields the distance function and its gradient simultaneously, with the distance component guaranteed to converge in $H^1(\Omega)$. Writing the network output as $F_\theta=d_\theta G_\theta$, with magnitude $d_\theta=\|F_\theta\|$ and unit direction $G_\theta=F_\theta/\|F_\theta\|$, the loss $\mathcal{L}_{\mathrm{NSP}}=\mathcal{L}_\Gamma+\lambda_{\mathrm{GM}}\mathcal{L}_{\mathrm{GM}}+\lambda_{\mathrm{SP}}\mathcal{L}_{\mathrm{SP}}$ enforces $d_\theta=0$ on the sampled surface, $\nabla d_\theta=G_\theta$, and $d_\theta(x-F_\theta(x))=0$, so that points pulled by the field land on its zero level set. The paper proves that a global minimizer of this loss is an exact shortest path to a surface containing the input samples, and that the magnitude converges in the $H^1$ norm as gradient matching is driven to zero.

Load-bearing premise

The theoretical guarantee applies to an idealized loss minimized exactly; the deployed training minimizes a modified objective with an extra minimal-area term and with one component of the shortest-path term kept fixed during optimization, and it is not proven that the modified objective has the same exact-shortest-path minimizer.

Editorial extensions

If this is right

Because the loss is built around the path $x-F_\theta(x)$ rather than around distances to the given points, the learned field is decoupled from the point cloud's quality; the paper cites this as the reason for improved robustness to noise and sparsity.
The single-network coupling of $d_\theta$ and $G_\theta$ yields a gradient field that is conservative by construction, avoiding the inconsistency the authors attribute to closest-surface-point methods that learn normals separately.
Convergence of $d_\theta$ in $H^1(\Omega)$ means the zero level set is controlled together with its gradient, reducing the kink-induced instability that appears when extracting meshes from distance functions.
The representation requires no inside or outside labels, so it applies to open and multi-layer surfaces such as garments, car interiors, and scanned indoor scenes where signed-distance or occupancy methods fail.

Reading between the lines

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

A testable extension is to check whether the $H^1$ guarantee survives the frozen outer gradient used in the reported shortest-path loss; proving convergence for that modified objective would require a different argument, likely a two-timescale or fixed-point analysis.
The same magnitude-direction decomposition could be applied to other unit-gradient field problems, such as geodesic distance estimation, fast marching, or level-set redistancing, where recovering distance and gradient together is equally desirable.
The minimal-area term is what separates true holes from missing scan regions; replacing it with a learned topology prior or a data-confidence mask would address the limitation the paper's conclusion identifies.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

This paper is worth reading. The core idea—parameterize F = d∇d as a single vector field, decompose it into magnitude and direction, and train with a shortest-path loss—is a real step beyond CSP, NDF, and CAP. It directly targets the gradient distortion at kinks that plagues distance-field networks, and the H1 convergence theorem (Theorem 1) is a clean conditional result: if the gradient-matching residual vanishes and the network converges, the magnitude converges in H1. The shortest-path surface extraction is also sensible and avoids relying on noisy gradients near the zero set. Empirically, the method is competitive or better on the reported open-surface benchmarks, with visually smoother outputs and qualitative robustness to noise and sparsity.

The soft spots are real but addressable. Most importantly, Theorem 2 proves that a global minimizer of LNSP (eq. 23) is an ESP, but training minimizes Ltotal (eq. 26), which adds a minimal-area term and detaches the gradient in the shortest-path loss. No theorem connects minimizers of Ltotal to the ESP characterization, and the detached gradient means the optimization trajectory is not even the gradient flow of LNSP. The abstract's claim that "its global minimum is the ESP" is therefore stronger than what is shown for the actual method. Second, the proof of Theorem 2 in Appendix A.2 applies a mean-value step to d*, which is only known to be in H1; without Lipschitz regularity along the path, the inequality |d*(a)-d*(b)| ≤ |a-b| needs justification. This is likely fixable, but the proof is incomplete as written. There is also a small gap in the Theorem 1 proof: an a.e. limit of H1 functions that are nonzero a.e. need not be nonzero a.e. The eikonal constraint probably saves the argument, but the step is not supplied. Finally, the "substantial improvements" wording overstates the tables—on cars and indoor scenes NSP is roughly on par with CAP—and the robustness comparisons are qualitative only, with no error bars or code.

Overall, the representation and the shortest-path loss are solid contributions, and the theory is close to being right but needs tightening. I would send this to peer review and ask for a revised version that either proves the global-minimum result for the loss actually minimized or clearly states the limitation, and that adds quantitative noise/sparsity metrics. I would cite it for the F = d∇d idea, not for its theorems until they are cleaned up.

Referee Report

4 major / 5 minor

Summary. The paper proposes the Neural Shortest Path (NSP), a vector-valued implicit neural representation Fθ = dθ Gθ whose magnitude approximates the unsigned distance function to a surface and whose unit direction approximates its gradient. The method decomposes the network output via magnitude-direction decomposition (MDD) and uses a variable-splitting training loss (manifold, gradient matching, and shortest-path terms) plus a minimal-area regularization. The authors claim two theoretical results: (i) Theorem 1 says that under a.e. convergence of the network output and vanishing gradient-matching loss, the magnitude converges in H1 norm; (ii) Theorem 2 says that a global minimizer of the idealized loss LNSP is the exact shortest path (ESP) for some surface Γ' containing the target Γ. They also introduce a shortest-path-based surface extraction algorithm and report experiments on synthetic, garment, car, indoor-scene, noisy, and sparse point clouds, showing qualitative and quantitative improvements over NDF, CAP, and CSP.

Significance. If the theoretical claims were fully established, the paper would make a valuable contribution: a single-network representation that simultaneously recovers distance and gradient, with a convergence guarantee in H1 avoiding higher-order derivatives, and a surface extraction method that bypasses the kinks at the zero level set. The empirical evaluation is extensive and suggests that the proposed loss is practically effective for open and non-watertight surfaces. Credit should be given for the explicit ESP construction, the variable-splitting formulation, and the reproducibility-oriented experimental reporting (datasets, baselines, and implementation details are given). However, the theoretical results as stated are consistency statements derived from loss terms designed to encode exactly the ESP properties; they do not certify that the trained model converges to the target surface. Moreover, the main theorem's proof has a regularity gap, and the theorem applies to a loss that is not the one minimized in the reported experiments. These issues are fixable in principle, but they currently leave the advertised guarantees without a direct link to the deployed method.

major comments (4)

[Theorem 1, Section 3.3 and Appendix A.1] Theorem 1 as stated is false: a sequence of nonzero a.e. H1 functions can converge a.e. to a limit that is zero on a set of positive measure (e.g., F_n = (1/n, 0, 0) on Ω), so the proof's assertion that 'the limit point F is also nonzero a.e.' is unjustified. The subsequent convergence proof for G_n = F_n/||F_n|| divides by d = ||F||, which is not well-defined on the zero set of the limit. The theorem needs an explicit assumption that the a.e.-limit F is nonzero a.e., or a separate handling of the zero set; as written, it does not establish the advertised H1 convergence.
[Theorem 2, Appendix A.2] The proof of Theorem 2 applies the multivariate mean-value theorem to d*, which is only known to lie in H1(Ω). In three dimensions, H1 functions are not necessarily continuous or differentiable at every point on the segment [x, x − d(x)∇d(x)], and the inequality |d*(a) − d*(b)| ≤ |a − b| does not follow from the weak relation |∇d*| = 1 a.e. without Lipschitz regularity or a viscosity/regularity argument. Thus the central comparison d*(x) ≤ d(x) and its reverse are not established. The proof needs an additional regularity statement for minimizers, e.g., showing that LSP = 0 plus LGM = 0 implies d* is the viscosity solution of the eikonal equation, not merely an H1 function.
[Section 3.3 vs. Section 3.2, Eq. (26) and (20)] The global-minimum statement is proved for LNSP in (23)/(24), but the reported models are trained with Ltotal (26), which adds the minimal-area term λMA ∫ δϵ(dθ) dx, and the shortest-path term is implemented with a detached outer gradient (20). No result shows that minimizers of Ltotal coincide with minimizers of LNSP or that the surrogate gradient in (20) preserves the global-minimum structure. Consequently, the theoretical guarantee does not directly apply to the objective actually minimized in the experiments, leaving the paper's central claim unsupported for the trained model.
[Theorem 2 statement and existence] Theorem 2 does not prove that the recovered surface equals the target surface: it only shows Γ' (the zero set of the minimizer) contains Γ, so the minimizer could be the distance function of any superset surface. The minimal-area term is thus essential for fidelity, but it is not part of the theoretical analysis. In addition, the paper assumes without proof that an optimal solution exists in the admissible set {F ∈ H1(Ω, R3) : F ≠ 0 a.e.}; the functional has no coercivity in ∇F itself and the nonzero constraint is not closed, so the infimum may not be attained. The theorem is vacuous unless existence is established.

minor comments (5)

[Section 3.3] There is a typo in the phrase 'demonstrate the the H1 norm convergence'; please remove the duplicated article.
[Section 3.2] The loss functional is introduced twice, once in Eq. (14) as the 'proposed loss functional' and again in Eq. (23) as LNSP with identical terms; please consolidate the notation to avoid confusion.
[Section 4.4] The noise standard deviations reported for the shirt data (σ = 0.002, 0.004, 0.005) are very small relative to the normalized domain [−1,1]^3; a brief note on the scale of the point-cloud coordinates would help the reader interpret the robustness results.
[Appendix B] There is a typo in the word 'utlizing' and the softplus formula would be clearer with explicit parentheses: σ(x) = (1/β) log(1 + e^(βx)).
[Section 3.2 and 3.3] The notation 'a.e.' is used extensively (e.g., Eq. (24), Theorem 1) before it is first defined; please define it at first use in Section 2.1 or 3.1.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the global-minimum theorem is a consistency result for a loss built from the defining ESP equations; no fitted parameter is renamed as a prediction and the empirical claims rest on external benchmarks.

full rationale

The only derivation with a self-definitional flavor is Theorem 2: the loss LNSP in (23) is a weighted sum of residuals of the defining equations of a distance function and of the ESP identity (boundary term LΓ, gradient-matching term LGM, and pullback term LSP), so proving that a global minimizer is the ESP is a consistency check that the zero-residual set of the loss is the intended object. This is by design rather than a fitted input being renamed as a prediction, and it does not feed back into the empirical comparisons. Theorem 1 is a conditional functional-analytic statement (a.e. convergence plus L2 convergence of ∇dn−Gn implies H1 convergence) with no fitted constants. The self-citations (Park et al. 2023/2024 for variable splitting, the similar-proof pointer, and the PINN remark) are not load-bearing: the proof of Theorem 2 is given in Appendix A.2, and the PINN remark is peripheral to the main derivation. Experiments are run against external baselines (NDF, CAP, CSP) with official codes on standard benchmarks, so the reported gains are not circularly defined. The main caveats—Theorem 2 is proved for LNSP rather than the Ltotal actually minimized, and the mean-value step in Appendix A.2 applies to an H1 function without established Lipschitz regularity—are correctness/rigor concerns, not circularity. Overall the circularity burden is low; score 2 reflects the minor self-citations and the by-construction flavor of the loss design.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The method introduces no invented physical entities; the exact shortest path is a mathematical definition, not a new force, particle, or conserved quantity. The theoretical results rest on standard Sobolev and eikonal facts plus several unproven regularity and implementation assumptions listed above. The free parameters are loss weights and extraction hyperparameters, including per-dataset λMA and an unreported δϵ width.

free parameters (5)

λSP = 0.01
Weight of the shortest-path loss; positive constant chosen by hand. The theorems hold for any positive value, but experiments use this fixed value.
λGM = 0.06
Weight of the gradient-matching loss; chosen by hand for all experiments.
λMA = 0.08, 0.15, 0.3 per dataset
Minimal-area regularization weight tuned separately for MGN, ShapeNet, and indoor scenes. This is per-dataset tuning that affects the reported reconstruction quality.
δϵ width ϵ = not reported
Width of the smeared Dirac delta in the minimal-area term is not given, so the actual regularizer used in experiments is not fully specified.
Extraction parameters = N=256, η=2/N, n=200, cell enlargement 7%
Hand-chosen parameters for the shortest-path surface extraction algorithm. They affect mesh quality but not the theoretical claim.

assumptions (6)

standard math Distance function to a bounded surface lies in H1(Ω) and satisfies |∇d| = 1 a.e.
Used in Sections 2.1 and 3.3 as the functional setting and eikonal property of the distance function.
domain assumption Ω is bounded, convex, simply connected and contains Γ, and the shortest path from x ∈ Ω to Γ stays in Ω.
Section 3 opening. Needed for the loss integrals over Ω and for the pulled point x - F(x) to remain in the domain where dθ is evaluated.
ad hoc to paper The network output Fθ is nonzero a.e. in Ω, so MDD and Gθ = Fθ/∥Fθ∥ are well defined.
Section 3.1 states this without a mechanism; softplus outputs can approach zero and division by the norm is a practical risk.
ad hoc to paper A global minimizer of LNSP exists in H1(Ω, R3) with F ≠ 0 a.e.
Equation (24) uses argmin, but no existence or attainment proof is given.
ad hoc to paper Theorem 2's comparison argument requires enough regularity of d* for a mean-value or integral step, and its zero set must be a set to which distance comparison applies.
Appendix A.2 invokes a mean-value property on the H1 minimizer without establishing regularity or the geometric nature of Γ′.
ad hoc to paper The detached-gradient implementation (20) has minima equivalent to the undetached LSP.
Section 3.2 describes the detach as an optimization device, but no statement or proof shows the two objectives share the same minimizers.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Neural Shortest Path for Surface Reconstruction from Point Clouds." pith.science (2026). https://pith.science/paper/UBZZIY5G

@misc{pith2026250206047,
  author       = {Pith},
  title        = {Pith review of: Neural Shortest Path for Surface Reconstruction from Point Clouds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UBZZIY5G}},
  note         = {Machine review of arXiv:2502.06047}
}

abstract

In this paper, we propose the neural shortest path (NSP), a vector-valued implicit neural representation (INR) that approximates a distance function and its gradient. The key feature of NSP is to learn the exact shortest path (ESP), which directs an arbitrary point to its nearest point on the target surface. The NSP is decomposed into its magnitude and direction, and a variable splitting method is used that each decomposed component approximates a distance function and its gradient, respectively. Unlike to existing methods of learning the distance function itself, the NSP ensures the simultaneous recovery of the distance function and its gradient. We mathematically prove that the decomposed representation of NSP guarantees the convergence of the magnitude of NSP in the $H^1$ norm. Furthermore, we devise a novel loss function that enforces the property of ESP, demonstrating that its global minimum is the ESP. We evaluate the performance of the NSP through comprehensive experiments on diverse datasets, validating its capacity to reconstruct high-quality surfaces with the robustness to noise and data sparsity. The numerical results show substantial improvements over state-of-the-art methods, highlighting the importance of learning the ESP, the product of distance function and its gradient, for representing a wide variety of complex surfaces.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 56 canonical work pages

[1]

ACM Transactions on Graphics (TOG) 32(2), 1–17 (2013)

Berger, M., Levine, J.A., Nonato, L.G., Taubin, G., Silva, C.T.: A benchmark for surface reconstruction. ACM Transactions on Graphics (TOG) 32(2), 1–17 (2013)

work page 2013
[2]

In: Computer Graphics Forum, vol

Sharf, A., Silva, C.T.: A survey of surface reconstruction from point clouds. In: Computer Graphics Forum, vol. 36, pp. 301–329 (2017). Wiley Online Library

work page 2017
[3]

IEEE transactions on visualization and computer graphics 5(4), 349–359 (1999)

Bernardini, F., Mittleman, J., Rushmeier, H., Silva, C., Taubin, G.: The ball- pivoting algorithm for surface reconstruction. IEEE transactions on visualization and computer graphics 5(4), 349–359 (1999)

work page 1999
[4]

In: Proceedings of the 28th Annual Conference on Computer Graphics and Interactive Techniques, pp

Carr, J.C., Beatson, R.K., Cherrie, J.B., Mitchell, T.J., Fright, W.R., McCallum, B.C., Evans, T.R.: Reconstruction and representation of 3d objects with radial basis 2https://github.com/jchibane/ndf 3https://github.com/junshengzhou/CAP-UDF 4https://github.com/rahulvenkk/csp-net 27 functions. In: Proceedings of the 28th Annual Conference on Computer Graph...

work page 2001
[5]

In: Proceedings IEEE Workshop on Variational and Level Set Methods in Computer Vision, pp

Zhao, H.-K., Osher, S., Fedkiw, R.: Fast surface reconstruction using the level set method. In: Proceedings IEEE Workshop on Variational and Level Set Methods in Computer Vision, pp. 194–201 (2001). IEEE

work page 2001
[6]

In: Pro- ceedings of the Eighth ACM Symposium on Solid Modeling and Applications, pp

Dey, T.K., Goswami, S.: Tight cocone: a water-tight surface reconstructor. In: Pro- ceedings of the Eighth ACM Symposium on Solid Modeling and Applications, pp. 127–134 (2003)

work page 2003
[7]

In: Proceedings of the Fourth Eurographics Symposium on Geometry Processing, vol

Kazhdan, M., Bolitho, M., Hoppe, H.: Poisson surface reconstruction. In: Proceedings of the Fourth Eurographics Symposium on Geometry Processing, vol. 7 (2006)

work page 2006
[8]

ACM Transactions on Graphics (ToG) 32(3), 1–13 (2013)

Kazhdan, M., Hoppe, H.: Screened poisson surface reconstruction. ACM Transactions on Graphics (ToG) 32(3), 1–13 (2013)

work page 2013

Show all 59 references

[9]

In: Computer Graphics Forum, vol

Digne, J., Morel, J.-M., Souzani, C.-M., Lartigue, C.: Scale space meshing of raw data point sets. In: Computer Graphics Forum, vol. 30, pp. 1630–1642 (2011). Wiley Online Library

2011
[10]

In: Proceedings of the 19th Annual Conference on Computer Graphics and Interactive Techniques, pp

Hoppe, H., DeRose, T., Duchamp, T., McDonald, J., Stuetzle, W.: Surface reconstruc- tion from unorganized points. In: Proceedings of the 19th Annual Conference on Computer Graphics and Interactive Techniques, pp. 71–78 (1992)

1992
[11]

In: Proceedings of the 21st Annual Conference on Computer Graphics and Interactive Techniques, pp

Hoppe, H., DeRose, T., Duchamp, T., Halstead, M., Jin, H., McDonald, J., Schweitzer, J., Stuetzle, W.: Piecewise smooth surface reconstruction. In: Proceedings of the 21st Annual Conference on Computer Graphics and Interactive Techniques, pp. 295–302 (1994)

1994
[12]

In: Proceedings of the 25th Annual Conference on Computer Graphics and Interactive Techniques, pp

Amenta, N., Bern, M., Kamvysselis, M.: A new voronoi-based surface reconstruction algorithm. In: Proceedings of the 25th Annual Conference on Computer Graphics and Interactive Techniques, pp. 415–421 (1998)

1998
[13]

In: Proceedings

Gopi, M., Krishnan, S.: A fast and efficient projection-based approach for surface reconstruction. In: Proceedings. XV Brazilian Symposium on Computer Graphics and Image Processing, pp. 179–186 (2002). IEEE

2002
[14]

In: Effective Computational Geometry for Curves and Surfaces, pp

Cazals, F., Giesen, J.: Delaunay triangulation based surface reconstruction. In: Effective Computational Geometry for Curves and Surfaces, pp. 231–276 (2006)

2006
[15]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Park, J.J., Florence, P., Straub, J., Newcombe, R., Lovegrove, S.: Deepsdf: Learning continuous signed distance functions for shape representation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 165–174 (2019)

2019
[16]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Mescheder, L., Oechsle, M., Niemeyer, M., Nowozin, S., Geiger, A.: Occupancy 28 networks: Learning 3d reconstruction in function space. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4460– 4470 (2019)

2019
[17]

In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part V, pp

Erler, P., Guerrero, P., Ohrhallinger, S., Mitra, N.J., Wimmer, M.: Points2surf learn- ing implicit surfaces from point clouds. In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part V, pp. 108–124 (2020). Springer

2020
[18]

International Conference on Neural Information Processing Systems (626) (2020)

Sitzmann, V., Martel, J.N.P., Bergman, A.W., Lindell, D.B., Wetzstein, G.: Implicit neural representations with periodic activation functions. International Conference on Neural Information Processing Systems (626) (2020)

2020
[19]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Zhang, J., Yao, Y., Li, S., Fang, T., McKinnon, D., Tsin, Y., Quan, L.: Criti- cal regularizations for neural surface reconstruction in the wild. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 6270–6279 (2022)

2022
[20]

In: Advances in Neural Information Processing Systems (2022)

Chou, G., Chugunov, I., Heide, F.: Gensdf: Two-stage learning of generalizable signed distance functions. In: Advances in Neural Information Processing Systems (2022)

2022
[21]

JMLR.org, 11 (2020)

Gropp, A., Yariv, L., Haim, N., Atzmon, M., Lipman, Y.: Implicit geometric regularization for learning shapes. JMLR.org, 11 (2020)

2020
[22]

In: International Conference on Machine Learning (2021)

Lipman, Y.: Phase transitions, distance functions, and implicit neural representations. In: International Conference on Machine Learning (2021)

2021
[23]

Advances in Neural Information Processing Systems 36 (2024)

Park, Y., Lee, T., Hahn, J., Kang, M.:p-poisson surface reconstruction in curl-free flow from point clouds. Advances in Neural Information Processing Systems 36 (2024)

2024
[24]

International Conference on Machine Learning (2021)

Ma, B., Han, Z., Liu, Y.-S., Zwicker, M.: Neural-pull: Learning signed distance func- tions from point clouds by learning to pull space onto surfaces. International Conference on Machine Learning (2021)

2021
[25]

arXiv preprint arXiv:1901.06802 (2019)

Michalkiewicz, M., Pontes, J.K., Jack, D., Baktashmotlagh, M., Eriksson, A.: Deep level sets: Implicit surface representations for 3d shape inference. arXiv preprint arXiv:1901.06802 (2019)

2019 arXiv
[26]

In: Computer Vision–ECCV 2020: 16th European Conference,

Peng, S., Niemeyer, M., Mescheder, L., Pollefeys, M., Geiger, A.: Convolutional occu- pancy networks. In: Computer Vision–ECCV 2020: 16th European Conference,

2020
[27]

523–540 (2020)

Glasgow, UK, August 23–28, 2020, Proceedings, Part III 16, pp. 523–540 (2020). Springer

2020
[28]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Mi, Z., Luo, Y., Tao, W.: Ssrnet: Scalable 3d surface reconstruction network. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 970–979 (2020) 29

2020
[29]

Advances in Neural Information Processing Systems 33, 21638–21652 (2020)

Chibane, J., Pons-Moll, G., et al.: Neural unsigned distance fields for implicit func- tion learning. Advances in Neural Information Processing Systems 33, 21638–21652 (2020)

2020
[30]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Chen, W., Lin, C., Li, W., Yang, B.: 3psdf: Three-pole signed distance function for learning surfaces with arbitrary topologies. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 18522–18531 (2022)

2022
[31]

In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

Zhou, J., Ma, B., Li, S., Liu, Y.-S., Han, Z.: Learning a more continuous zero level set in unsigned distance fields through level set projection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 3181–3192 (2023)

2023
[32]

In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

Zhao, F., Wang, W., Liao, S., Shao, L.: Learning anchored unsigned distance func- tions with gradient direction alignment for single-view garment reconstruction. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 12674–12683 (2021)

2021
[33]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Long, X., Lin, C., Liu, L., Liu, Y., Wang, P., Theobalt, C., Komura, T., Wang, W.: Neuraludf: Learning unsigned distance fields for multi-view reconstruction of sur- faces with arbitrary topologies. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Reco...

2023
[34]

In: European Conference on Computer Vision, pp

Guillard, B., Stella, F., Fua, P.: Meshudf: Fast and differentiable meshing of unsigned distance field networks. In: European Conference on Computer Vision, pp. 576–592 (2022). Springer

2022
[35]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

Ye, J., Chen, Y., Wang, N., Wang, X.: GIFS: Neural implicit function for general shape representation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12829–12839 (2022)

2022
[36]

In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

Venkatesh, R., Karmali, T., Sharma, S., Ghosh, A., Babu, R.V., Jeni, L.A., Singh, M.: Deep implicit surface point prediction networks. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 12653–12662 (2021)

2021
[37]

IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

Zhou, J., Ma, B., Li, S., Liu, Y.-S., Fang, Y., Han, Z.: Cap-udf: Learning unsigned distance functions progressively from raw point clouds with consistency-aware field optimization. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

2024
[38]

Elsevier (2003)

Adams, R.A., Fournier, J.J.: Sobolev spaces. Elsevier (2003)

2003
[39]

arXiv preprint arXiv:2305.08174 (2023)

Park, Y., Song, C.H., Hahn, J., Kang, M.: Resdf: Redistancing implicit surfaces using neural networks. arXiv preprint arXiv:2305.08174 (2023)

2023 arXiv
[40]

ACM SIGGRAPH computer graphics 21(4), 163–169 (1987) 30

Lorensen, W.E., Cline, H.E.: Marching cubes: A high resolution 3D surface construc- tion algorithm. ACM SIGGRAPH computer graphics 21(4), 163–169 (1987) 30

1987
[41]

ACM Transactions on Graphics (TOG) 41(4), 1–13 (2022)

Chen, Z., Tagliasacchi, A., Funkhouser, T., Zhang, H.: Neural dual contouring. ACM Transactions on Graphics (TOG) 41(4), 1–13 (2022)

2022
[42]

In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

Zhang, C., Lin, G., Yang, L., Li, X., Komura, T., Schaefer, S., Keyser, J., Wang, W.: Surface extraction from neural unsigned distance fields. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 22531–22540 (2023)

2023
[43]

ACM Transactions on Graphics (TOG) 42(6), 1–15 (2023)

Hou, F., Chen, X., Wang, W., Qin, H., He, Y.: Robust zero level-set extraction from unsigned distance fields based on double covering. ACM Transactions on Graphics (TOG) 42(6), 1–15 (2023)

2023
[44]

In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp

Bhatnagar, B.L., Tiwari, G., Theobalt, C., Pons-Moll, G.: Multi-garment net: Learning to dress 3d people from images. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 5420–5430 (2019)

2019
[45]

Technical Report arXiv:1512.03012 [cs.GR], Stanford University — Princeton University — Toyota Technological Institute at Chicago (2015)

Savva, M., Song, S., Su, H., Xiao, J., Yi, L., Yu, F.: Shapenet: An information-rich 3d model repository. Technical Report arXiv:1512.03012 [cs.GR], Stanford University — Princeton University — Toyota Technological Institute at Chicago (2015)

2015 arXiv
[46]

ACM Transactions on Graphics (ToG) 32(4v), 1–8 (2013)

Zhou, Q.-Y., Koltun, V.: Dense scene reconstruction with points of interest. ACM Transactions on Graphics (ToG) 32(4v), 1–8 (2013)

2013
[47]

American Mathematical Society 19 (2022)

Evans, L.C.: Partial differential equations. American Mathematical Society 19 (2022)

2022
[48]

Desmaison, A., Antiga, L., Lerer, A.: Automatic differentiation in pytorch (2017)

2017
[49]

Advances in neural information processing systems 30 (2017)

Czarnecki, W.M., Osindero, S., Jaderberg, M., Swirszcz, G., Pascanu, R.: Sobolev training for neural networks. Advances in neural information processing systems 30 (2017)

2017
[50]

arXiv preprint arXiv:2101.08932 (2021)

Son, H., Jang, J.W., Han, W.J., Hwang, H.J.: Sobolev training for physics informed neural networks. arXiv preprint arXiv:2101.08932 (2021)

2021 arXiv
[51]

Advances in Neural Information Processing Systems (2023)

Yang, H., Sun, Y., Sundaramoorthi, G., Yezzi, A.: Steik: Stabilizing the optimization of neural signed distance functions and finer shape representation. Advances in Neural Information Processing Systems (2023)

2023
[52]

Advances in Neural Information Processing Systems 36 (2024)

Wang, R., Wang, Z., Zhang, Y., Chen, S., Xin, S., Tu, C., Wang, W.: Aligning gradient and hessian for neural signed distance function. Advances in Neural Information Processing Systems 36 (2024)

2024
[53]

arXiv preprint arXiv:2409.20383 (2024)

Park, Y., Song, C., Kang, M.: Beyond derivative pathology of pinns: Variable splitting strategy with convergence analysis. arXiv preprint arXiv:2409.20383 (2024)

2024 arXiv
[54]

Communications of the ACM 20(10), 693–702 (1977) 31

Fuchs, H., Kedem, Z.M., Uselton, S.P.: Optimal surface reconstruction from planar contours. Communications of the ACM 20(10), 693–702 (1977) 31

1977
[55]

In: Advances in Neural Information Processing Systems (2022)

Pumarola, A., Sanakoyeu, A., Yariv, L., Thabet, A.K., Lipman, Y.: Visco grids: Surface reconstruction with viscosity and coarea grids. In: Advances in Neural Information Processing Systems (2022)

2022
[56]

Computer Vision and Image Understanding 111(3), 351– 373 (2008)

Caselles, V., Haro, G., Sapiro, G., Verdera, J.: On geometric variational models for inpainting surface holes. Computer Vision and Image Understanding 111(3), 351– 373 (2008)

2008
[57]

SIAM Journal on Imaging Sciences 13(4), 1834–1859 (2020)

He, Y., Kang, S.H., Liu, H.: Curvature regularized surface reconstruction from point clouds. SIAM Journal on Imaging Sciences 13(4), 1834–1859 (2020)

2020
[58]

In: Proceedings of the 29th Annual Conference on Computer Graphics and Interactive Techniques, pp

Ju, T., Losasso, F., Schaefer, S., Warren, J.: Dual contouring of hermite data. In: Proceedings of the 29th Annual Conference on Computer Graphics and Interactive Techniques, pp. 339–346 (2002)

2002
[59]

arXiv preprint arXiv:1412.6980 (2014) 32

Kingma, D.P., Ba, J.: Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014) 32

2014 arXiv

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [1]

ACM Transactions on Graphics (TOG) 32(2), 1–17 (2013)

Berger, M., Levine, J.A., Nonato, L.G., Taubin, G., Silva, C.T.: A benchmark for surface reconstruction. ACM Transactions on Graphics (TOG) 32(2), 1–17 (2013)

work page 2013

[2] [2]

In: Computer Graphics Forum, vol

Sharf, A., Silva, C.T.: A survey of surface reconstruction from point clouds. In: Computer Graphics Forum, vol. 36, pp. 301–329 (2017). Wiley Online Library

work page 2017

[3] [3]

IEEE transactions on visualization and computer graphics 5(4), 349–359 (1999)

Bernardini, F., Mittleman, J., Rushmeier, H., Silva, C., Taubin, G.: The ball- pivoting algorithm for surface reconstruction. IEEE transactions on visualization and computer graphics 5(4), 349–359 (1999)

work page 1999

[4] [4]

In: Proceedings of the 28th Annual Conference on Computer Graphics and Interactive Techniques, pp

Carr, J.C., Beatson, R.K., Cherrie, J.B., Mitchell, T.J., Fright, W.R., McCallum, B.C., Evans, T.R.: Reconstruction and representation of 3d objects with radial basis 2https://github.com/jchibane/ndf 3https://github.com/junshengzhou/CAP-UDF 4https://github.com/rahulvenkk/csp-net 27 functions. In: Proceedings of the 28th Annual Conference on Computer Graph...

work page 2001

[5] [5]

In: Proceedings IEEE Workshop on Variational and Level Set Methods in Computer Vision, pp

Zhao, H.-K., Osher, S., Fedkiw, R.: Fast surface reconstruction using the level set method. In: Proceedings IEEE Workshop on Variational and Level Set Methods in Computer Vision, pp. 194–201 (2001). IEEE

work page 2001

[6] [6]

In: Pro- ceedings of the Eighth ACM Symposium on Solid Modeling and Applications, pp

Dey, T.K., Goswami, S.: Tight cocone: a water-tight surface reconstructor. In: Pro- ceedings of the Eighth ACM Symposium on Solid Modeling and Applications, pp. 127–134 (2003)

work page 2003

[7] [7]

In: Proceedings of the Fourth Eurographics Symposium on Geometry Processing, vol

Kazhdan, M., Bolitho, M., Hoppe, H.: Poisson surface reconstruction. In: Proceedings of the Fourth Eurographics Symposium on Geometry Processing, vol. 7 (2006)

work page 2006

[8] [8]

ACM Transactions on Graphics (ToG) 32(3), 1–13 (2013)

Kazhdan, M., Hoppe, H.: Screened poisson surface reconstruction. ACM Transactions on Graphics (ToG) 32(3), 1–13 (2013)

work page 2013