pith. sign in

arxiv: 2607.00492 · v1 · pith:N4BIFTTZnew · submitted 2026-07-01 · 💻 cs.CV

GenSP: Consistent Spherical Parameterization via Learning Shape Generative Models

Sai Karthikey Pentapati , Shashank Gupta , Rajesh Sureddi , Yuezhi Yang , Alan C. Bovik , Qixing Huang This is my paper

Pith reviewed 2026-07-02 14:56 UTC · model grok-4.3

classification 💻 cs.CV
keywords spherical parameterizationgenerative modelneural networkshape deformationconsistent mappinggenus-0 shapesShapeNet dataset
0
0 comments X

The pith

Learning a neural generative model from the unit sphere to shapes produces consistent spherical parameterizations by using inverse mappings.

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

The paper presents GenSP, a framework that learns a generative model mapping the unit sphere to a collection of genus-0 shapes. Spherical parameterizations come from the inverse of this mapping, which naturally encourages consistency between similar shapes. This is achieved with a continuous neural deformation model, augmented training with bridging shapes, and initial correspondences from a spanning tree in latent space. The result is lower distortion and higher consistency than optimizing each shape separately, as shown on ShapeNet data.

Core claim

By training a neural network to map sphere coordinates and latent codes to surface points on shapes, the inverse operation yields spherical parameterizations that are consistent across the dataset because the model learns shared deformations.

What carries the argument

Continuous neural deformation model predicting surface points from sphere coordinates and latent shape codes, with inverse mappings providing the parameterizations.

If this is right

  • Similar shapes in the collection receive aligned spherical parameterizations.
  • Geometric distortion is reduced compared to independent per-shape optimization.
  • Cross-shape consistency improves without requiring explicit correspondence optimization.
  • The method handles heterogeneous collections through intermediate bridging shapes.

Where Pith is reading between the lines

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

  • This consistency could improve downstream tasks such as texture transfer or shape interpolation across the dataset.
  • The generative approach might allow parameterizing new shapes not in the training set by using the learned model.
  • Extending the method to non-genus-0 shapes would require handling topological variations.

Load-bearing premise

The continuous neural deformation model combined with intermediate bridging shapes and spanning-tree initial correspondences can learn meaningful deformations across a heterogeneous collection without introducing inconsistencies or artifacts.

What would settle it

A test where two very similar shapes receive spherical parameterizations that map corresponding surface features to widely different sphere locations would falsify the consistency benefit.

Figures

Figures reproduced from arXiv: 2607.00492 by Alan C. Bovik, Qixing Huang, Rajesh Sureddi, Sai Karthikey Pentapati, Shashank Gupta, Yuezhi Yang.

Figure 1
Figure 1. Figure 1: Comparison of spherical parameterizations with existing approaches. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: An overview of the four stages of GenSP. trapped in local minima. To address this challenge, we propose to introduce an implicit generative model f ϕ : R 3 × R d → R to facilitate the computation of initial correspondences for learning this shape generative model. As shown in [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (Top) If we deform S0 to align Si through intermediate shapes defined by f ϕ (·, z) the alignment is poor, as some intermediate shapes have large distortions. (Bottom) When we deform S0 to align Si through intermediate shapes between adjacent training shapes on a spanning tree rooted at S0, the alignment is more accurate as intermediate shapes are much more meaningful. distribution of shapes in the AE late… view at source ↗
Figure 4
Figure 4. Figure 4: Comparisons between (Top) initial correspondences between S0 and each train￾ing shape and (Bottom) the correspondences obtained by optimized shape generator. We can see that the optimized correspondences are smoother and align the training data (green) better than the initial correspondences. This is attributed to the defor￾mation loss on the mesh generator, which optimizes all correspondences together. Le… view at source ↗
Figure 5
Figure 5. Figure 5: Additional qualitative spherical parameterization results. The top row in each block visualizes the mapping of each method from a shape to the sphere. (The spheres are manually rotated for ease of visualization.) The bottom row shows the mapping of each method from the sphere to the shape. We can see that our method offers more consistent and high-quality parameterizations. Source GenSP Diff3F DiffumatchUL… view at source ↗
Figure 6
Figure 6. Figure 6: Comparison against modern correspon￾dence methods. Method Bad # Affected face % meshes SMAT 0.004 4 AHSP 0.002 3 CMCF 0.241 15 ARAP 0.434 31 VC25 1.898 91 GenSP 0.099 9 [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Application in inter-shape correspondences. Baselines are Mobius registration (MR), blended intrinsic maps (BIM), and elastic functional maps (EFM). Note that EFM needs an initial set of correspondences for initialization, which is why it fails in a zero-shot setting. D Limitations All methods of spherical parameterization, including ours and the baselines shown in this paper, fail for shapes with thin str… view at source ↗
Figure 8
Figure 8. Figure 8: More qualitative comparisons between GenSP and CMCF [33], SMAT [50],and AHSP [28]. We show images of the same sphere texture under different spherical pa￾rameterizations [PITH_FULL_IMAGE:figures/full_fig_p027_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative ablation study results [PITH_FULL_IMAGE:figures/full_fig_p028_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Application in shape morphing. Applying GenSP’s decoder on interpolated latent vectors produces smoothly morphed shapes [PITH_FULL_IMAGE:figures/full_fig_p028_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: An example of GenSP’s failure mode [PITH_FULL_IMAGE:figures/full_fig_p029_11.png] view at source ↗
read the original abstract

We introduce GenSP, a data-driven framework that learns consistent spherical parameterizations across a collection of genus-0 shapes. Instead of optimizing the parameterization of each shape independently, our method learns a neural generative model that predicts a continuous mapping from the unit sphere to shapes in a dataset. Under this formulation, spherical parameterizations are obtained through the inverse mappings of the learned generator, which encourages similar shapes to share consistent parameterizations. To make this formulation practical, we address several key challenges in learning such a generative model. First, we introduce a continuous neural deformation model that predicts surface points from sphere coordinates and latent shape codes, avoiding discretization artifacts common in mesh-based formulations. Second, we augment the training space with intermediate shapes that bridge the sphere and input shapes, allowing the model to learn meaningful deformations across a heterogeneous shape collection. Third, we compute reliable initial correspondences by propagating mappings along a spanning tree of training shapes in the latent space. Experiments on the ShapeNet dataset demonstrate that our approach significantly reduces geometric distortion and improves cross-shape consistency compared with state-of-the-art spherical parameterization methods.

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.

Referee Report

2 major / 2 minor

Summary. The paper introduces GenSP, a data-driven approach to consistent spherical parameterization of genus-0 shapes. Rather than optimizing each shape independently, it trains a neural generative model G(u, z) that maps unit-sphere coordinates u and latent shape codes z to surface points; spherical parameterizations are recovered by inverting the per-shape map u ↦ G(u, z_shape). Practicality is achieved via a continuous neural deformation model, augmentation with intermediate bridging shapes, and initial correspondences obtained by propagating mappings along a spanning tree in latent space. Experiments on ShapeNet are reported to show lower geometric distortion and higher cross-shape consistency than prior spherical parameterization methods.

Significance. If the learned mappings are reliably bijective and the consistency gains hold under quantitative verification, the shift from per-shape optimization to a shared generative model would be a meaningful contribution to shape correspondence and parameterization tasks in computer vision and graphics. The spanning-tree initialization and bridging-shape augmentation are concrete technical ideas that could be reused even if the overall framework requires refinement.

major comments (2)
  1. [Abstract / method formulation] Abstract and the description of the continuous neural deformation model: the central claim that parameterizations are obtained via inverse mappings of G(u, z) requires the per-shape map to be bijective. No Jacobian-determinant penalty, cycle-consistency term, or diffeomorphic constraint is stated, leaving open the possibility that the optimization produces folds or uncovered regions on heterogeneous ShapeNet collections; this directly undermines the consistency guarantee.
  2. [method (spanning-tree step)] The spanning-tree propagation of initial correspondences (third key challenge) is presented as auxiliary, yet the consistency of the final learned inverses depends on the quality of these seeds. Without an ablation that isolates the tree construction or measures how many shapes receive poor initial maps, it is unclear whether the reported consistency improvements are robust or an artifact of favorable tree ordering.
minor comments (2)
  1. [Abstract] The abstract claims 'significantly reduces geometric distortion' but supplies no numerical values, baseline names, or table references; readers cannot assess effect size from the provided text.
  2. [Abstract / method] Notation for the generator G(u, z) and the inverse operation is introduced without an explicit equation or diagram showing how the inverse is computed at inference time (e.g., optimization-based inversion versus an auxiliary decoder).

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback highlighting the importance of bijectivity guarantees and the robustness of the spanning-tree initialization. We address each major comment below and propose targeted revisions to strengthen the claims.

read point-by-point responses

  1. Referee: [Abstract / method formulation] Abstract and the description of the continuous neural deformation model: the central claim that parameterizations are obtained via inverse mappings of G(u, z) requires the per-shape map to be bijective. No Jacobian-determinant penalty, cycle-consistency term, or diffeomorphic constraint is stated, leaving open the possibility that the optimization produces folds or uncovered regions on heterogeneous ShapeNet collections; this directly undermines the consistency guarantee.

    Authors: We acknowledge that the abstract does not explicitly mention bijectivity constraints. The continuous neural deformation model, bridging-shape augmentation, and training losses are intended to encourage smooth, invertible mappings through the generative formulation, but we agree this is not rigorously justified in the provided description. In the revision we will expand the method section with a dedicated discussion of bijectivity, add a Jacobian-determinant regularization term to the objective, and report quantitative checks (e.g., sign of the determinant and coverage metrics) on the learned inverses. revision: yes

  2. Referee: [method (spanning-tree step)] The spanning-tree propagation of initial correspondences (third key challenge) is presented as auxiliary, yet the consistency of the final learned inverses depends on the quality of these seeds. Without an ablation that isolates the tree construction or measures how many shapes receive poor initial maps, it is unclear whether the reported consistency improvements are robust or an artifact of favorable tree ordering.

    Authors: The spanning-tree step supplies the initial correspondences required for the generative model to learn consistent inverses; we therefore agree it is not merely auxiliary. The current manuscript does not contain an ablation isolating its contribution or statistics on seed quality. In the revision we will add an ablation study that varies the tree construction (e.g., random vs. latent-space MST) and report the fraction of shapes whose initial maps exhibit high distortion, thereby quantifying robustness. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained

full rationale

The paper defines spherical parameterizations explicitly as the inverse mappings of a learned generative model G(u, z) from unit sphere coordinates u and latent codes z to surface points. This is a methodological choice rather than a derivation that reduces to its own inputs by construction. Training augments the space with bridging shapes and uses spanning-tree initial correspondences as auxiliary steps to enable learning across heterogeneous data, but these do not create a self-referential loop where a claimed prediction equals a fitted input. No self-citations are invoked as load-bearing uniqueness theorems, no ansatz is smuggled, and no known empirical pattern is merely renamed. The central claim of improved consistency via shared latent space remains independent of the parameterization definition itself.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review; specific free parameters, axioms, and invented entities cannot be enumerated from the provided text.

axioms (1)
  • domain assumption All input shapes are genus-0 surfaces that admit spherical parameterization.
    Stated in the abstract as the target class of shapes.

pith-pipeline@v0.9.1-grok · 5739 in / 1109 out tokens · 23676 ms · 2026-07-02T14:56:11.248484+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

71 extracted references · 47 canonical work pages · 1 internal anchor

  1. [1]

    ACM Trans

    Aigerman, N., Gupta, K., Kim, V.G., Chaudhuri, S., Saito, J., Groueix, T.: Neural jacobian fields: learning intrinsic mappings of arbitrary meshes. ACM Trans. Graph. 41(4) (Jul 2022). https://doi.org/10.1145/3528223.3530141 , https://doi.org/10.1145/3528223.3530141

    work page doi:10.1145/3528223.3530141 2022

  2. [2]

    The Visual Com- puter 16(1), 26–37 (2000).https://doi.org/10.1007/s003710050003, doi.org

    Alexa, M.: Merging polyhedral shapes with scattered features. The Visual Com- puter 16(1), 26–37 (2000).https://doi.org/10.1007/s003710050003, doi.org

    work page doi:10.1007/s003710050003 2000

  3. [3]

    In: International Conference on 3D Vision, 3DV 2021, London, United Kingdom, De- cember 1-3, 2021

    Attaiki, S., Pai, G., Ovsjanikov, M.: DPFM: deep partial functional maps. In: International Conference on 3D Vision, 3DV 2021, London, United Kingdom, De- cember 1-3, 2021. pp. 175–185. IEEE, London, United Kingdom (2021).https:// doi.org/10.1109/3DV53792.2021.00040, https://doi.org/10.1109/3DV53792. 2021.00040

    work page doi:10.1109/3dv53792.2021.00040 2021

  4. [4]

    In: 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021

    Atzmon, M., Lipman, Y.: SALD: sign agnostic learning with derivatives. In: 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021. pp. –. OpenReview.net, Virtual Event, Austria (2021), https://openreview.net/forum?id=7EDgLu9reQD

    2021

  5. [5]

    Atzmon, M., Novotný, D., Vedaldi, A., Lipman, Y.: Augmenting implicit neural shape representations with explicit deformation fields (2021),https://arxiv.org/ abs/2108.08931

    work page arXiv 2021

  6. [6]

    Baden, A., Crane, K., Kazhdan, M.: Möbius registration. Comput. Graph. Forum 37(5), 211–220 (2018).https://doi.org/10.1111/CGF.13503, https://doi.org/ 10.1111/cgf.13503

    work page doi:10.1111/cgf.13503 2018

  7. [7]

    In: Computer Graphics Forum

    Baden, A., Crane, K., Kazhdan, M.: Möbius registration. In: Computer Graphics Forum. vol. 37, pp. 211–220. Wiley Online Library (2018)

    2018

  8. [8]

    In: 2019 IEEE/CVF International Conference on Com- puter Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019

    Bouritsas, G., Bokhnyak, S., Ploumpis, S., Zafeiriou, S., Bronstein, M.M.: Neural 3d morphable models: Spiral convolutional networks for 3d shape representation learning and generation. In: 2019 IEEE/CVF International Conference on Com- puter Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019. pp. 7212–7221. IEEE, New York, NY, USA (20...

    work page arXiv 2019

  9. [9]

    ACM TOG (2023)

    Cao, D., Roetzer, P., Bernard, F.: Unsupervised learning of robust spectral shape matching. ACM TOG (2023)

    2023

  10. [10]

    ShapeNet: An Information-Rich 3D Model Repository

    Chang, A.X., Funkhouser, T., Guibas, L., Hanrahan, P., Huang, Q., Li, Z., Savarese, S., Savva, M., Song, S., Su, H., et al.: Shapenet: An information-rich 3d model repository. arXiv preprint arXiv:1512.03012 (2015)

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  11. [11]

    Chen, W., Xiao, D., Chen, R.: Bijective spherical parameterization via stereo- graphic projection. Vis. Comput.41(9), 6413–6424 (2025).https://doi.org/10. 1007/S00371-025-03942-5, https://doi.org/10.1007/s00371-025-03942-5

    work page doi:10.1007/s00371-025-03942-5 2025

  12. [13]

    In: Vedaldi, A., Bischof, H., Brox, T., Frahm, J

    Cosmo, L., Norelli, A., Halimi, O., Kimmel, R., Rodolà, E.: LIMP: learning latent shape representations with metric preservation priors. In: Vedaldi, A., Bischof, H., Brox, T., Frahm, J. (eds.) Computer Vision - ECCV 2020 - 16th European GenSP: Consistent Spherical Parameterizations 17 Conference,Glasgow,UK,August23-28,2020,Proceedings,PartIII.LectureNote...

    work page doi:10.1007/978-3-030-58580-8_2 2020

  13. [14]

    ACM Trans

    Crane, K., Pinkall, U., Schröder, P.: Robust fairing via conformal curvature flow. ACM Trans. Graph. 32(4) (Jul 2013). https://doi.org/10.1145/2461912. 2461986, https://doi.org/10.1145/2461912.2461986

    work page doi:10.1145/2461912 2013

  14. [15]

    ACM Trans

    Crane, K., Pinkall, U., Schröder, P.: Robust fairing via conformal curvature flow. ACM Trans. Graph.32 (2013)

    2013

  15. [16]

    Das, G., Goodrich, M.T.: On the complexity of optimization problems for 3- dimensional convex polyhedra and decision trees. Comput. Geom. Theory Appl. 8(3), 123–137 (1997). https : / / doi . org / 10 . 1016 / S0925 - 7721(97 ) 00006 - 0, https://doi.org/10.1016/S0925-7721(97)00006-0

    work page doi:10.1016/s0925-7721(97)00006-0 1997

  16. [17]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp 5891–5900, https://doi.org/10

    Donati, N., Corman, E., Ovsjanikov, M.: Deep orientation-aware functional maps: Tackling symmetry issues in shape matching. In: IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022. pp. 732–741. IEEE, Washington, DC, USA (2022).https://doi.org/ 10.1109/CVPR52688.2022.00082 , https://doi.org/10...

    work page doi:10.1109/cvpr52688.2022.00082 2022

  17. [18]

    ALFRED: A benchmark for interpreting grounded instructions for everyday tasks

    Donati, N., Sharma, A., Ovsjanikov, M.: Deep geometric functional maps: Ro- bust feature learning for shape correspondence. In: 2020 IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, CVPR 2020, Seattle, WA, USA, June 13-19, 2020. pp. 8589–8598. Computer Vision Foundation / IEEE, Washington, DC, USA (2020). https://doi.org/10.1109/CVPR42600...

    work page doi:10.1109/cvpr42600.2020 2020

  18. [19]

    In: IEEE CVPR (2024)

    Dutt, N.S., Muralikrishnan, S., Mitra, N.J.: Diffusion 3d features (diff3f): Deco- rating untextured shapes with distilled semantic features. In: IEEE CVPR (2024)

    2024

  19. [20]

    Eisenberger, M., Lähner, Z., Cremers, D.: Divergence-free shape correspondence by deformation. Comput. Graph. Forum38(5), 1–12 (2019).https://doi.org/ 10.1111/CGF.13785, https://doi.org/10.1111/cgf.13785

    work page doi:10.1111/cgf.13785 2019

  20. [21]

    In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021

    Eisenberger, M., Novotný, D., Kerchenbaum, G., Labatut, P., Neverova, N., Cremers, D., Vedaldi, A.: Neuromorph: Unsupervised shape interpolation and correspondence in one go. In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021. pp. 7473–7483. Computer Vision Foundation / IEEE, Washington, DC, USA (2021). ht...

    2021

  21. [22]

    In: ACM SIGGRAPH 2003 Papers

    Gotsman, C., Gu, X., Sheffer, A.: Fundamentals of spherical parameterization for 3d meshes. In: ACM SIGGRAPH 2003 Papers. p. 358–363. SIGGRAPH ’03, As- sociation for Computing Machinery, New York, NY, USA (2003).https://doi. org/10.1145/1201775.882276, https://doi.org/10.1145/1201775.882276

    work page doi:10.1145/1201775.882276 2003

  22. [23]

    In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y

    Groueix, T., Fisher, M., Kim, V.G., Russell, B.C., Aubry, M.: 3d-coded: 3d corre- spondences by deep deformation. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) Computer Vision - ECCV 2018 - 15th European Conference, Munich, Germany, September 8-14, 2018, Proceedings, Part II. Lecture Notes in Computer Science, vol. 11206, pp. 235–251. Sp...

    work page doi:10.1007/978-3-030-01216-8_15 2018

  23. [24]

    IEEE Transactions on Visualization and Computer Graphics (TVCG) 6(2), 181–189 (2000)

    Haker, S., Angenent, S., Tannenbaum, A., Kikinis, R., Sapiro, G., Halle, M.: Conformal surface parameterization for texture mapping. IEEE Transactions on Visualization and Computer Graphics (TVCG) 6(2), 181–189 (2000). https: //doi.org/10.1109/2945.856994

    work page doi:10.1109/2945.856994 2000

  24. [25]

    In: ACM SIGGRAPH 2023 Conference Pro- ceedings

    Hartwig, F., Sassen, J., Azencot, O., Rumpf, M., Ben-Chen, M.: An elastic basis for spectral shape correspondence. In: ACM SIGGRAPH 2023 Conference Pro- ceedings. SIGGRAPH ’23, Association for Computing Machinery, New York, NY, USA (2023). https://doi.org/10.1145/3588432.3591518 , https://doi.org/ 10.1145/3588432.3591518

    work page doi:10.1145/3588432.3591518 2023

  25. [26]

    In: ACM SIGGRAPH 2023 conference proceedings

    Hartwig, F., Sassen, J., Azencot, O., Rumpf, M., Ben-Chen, M.: An elastic basis for spectral shape correspondence. In: ACM SIGGRAPH 2023 conference proceedings. pp. 1–11 (2023)

    2023

  26. [27]

    In: McMains, S., Sloan, P

    Hormann, K., Lévy, B., Sheffer, A.: Mesh parameterization: theory and practice video files associated with this course are available from the citation page. In: McMains, S., Sloan, P. (eds.) International Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2007, San Diego, California, USA, August 5- 9, 2007, Courses. p. 1. ACM, San Diego,...

    work page doi:10.1145/1281500.1281510 2007

  27. [28]

    IEEE Transactions on Visualization and Computer Graphics24(6), 1930–1941 (2017)

    Hu, X., Fu, X.M., Liu, L.: Advanced hierarchical spherical parameterizations. IEEE Transactions on Visualization and Computer Graphics24(6), 1930–1941 (2017)

    1930

  28. [30]

    Frozen in time: A joint video and image encoder for end-to-end retrieval

    Huang, Q., Huang, X., Sun, B., Zhang, Z., Jiang, J., Bajaj, C.: Arapreg: An as-rigid-as possible regularization loss for learning deformable shape generators. In: 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021, Montreal, QC, Canada, October 10-17, 2021. pp. 5795–5805. IEEE, New York, NY, USA (2021). https://doi.org/10.1109/ICCV48922....

    work page doi:10.1109/iccv48922.2021.00576 2021

  29. [31]

    ACM Trans

    Huang, Q., Wang, F., Guibas, L.J.: Functional map networks for analyzing and exploring large shape collections. ACM Trans. Graph.33(4), 36:1–36:11 (2014), https://doi.org/10.1145/2601097.2601111

    work page doi:10.1145/2601097.2601111 2014

  30. [32]

    Computer Graphics Forum28(2), 407–416 (2009).https://doi.org/10

    Huang, Q., Wicke, M., Adams, B., Guibas, L.: Shape decomposition using modal analysis. Computer Graphics Forum28(2), 407–416 (2009).https://doi.org/10. 1111/j.1467-8659.2009.01380.x

    work page arXiv 2009

  31. [33]

    Kazhdan, M., Solomon, J., Ben-Chen, M.: Can mean-curvature flow be modified to be non-singular? Comput. Graph. Forum31(5), 1745–1754 (2012).https://doi. org/10.1111/J.1467-8659.2012.03179.X , https://doi.org/10.1111/j.1467- 8659.2012.03179.x

    work page doi:10.1111/j.1467-8659.2012.03179.x 2012

  32. [34]

    ACM Trans

    Kim, V.G., Li, W., Mitra, N.J., DiVerdi, S., Funkhouser, T.: Exploring collections of 3d models using fuzzy correspondences. ACM Trans. Graph.31(4), 54:1–54:11 (Jul 2012), 10.1145/2185520.2185550

    work page doi:10.1145/2185520.2185550 2012

  33. [35]

    Kim,V.G.,Lipman,Y.,Funkhouser,T.:Blendedintrinsicmaps.ACMtransactions on graphics (TOG)30(4), 1–12 (2011)

    2011

  34. [36]

    Computer Graphics Forum18(3), 119–130 (1999)

    Kobbelt, L.P., Vorsatz, J., Labsik, U., Seidel, H.P.: A shrink-wrapping approach to remeshing polygonal surfaces. Computer Graphics Forum18(3), 119–130 (1999). https://doi.org/10.1111/1467-8659.00319, dl.acm.org GenSP: Consistent Spherical Parameterizations 19

    work page doi:10.1111/1467-8659.00319 1999

  35. [37]

    ACM Trans

    Lipman, Y., Funkhouser, T.: Möbius voting for surface correspondence. ACM Trans. Graph. 28(3) (Jul 2009). https://doi.org/10.1145/1531326.1531378 , https://doi.org/10.1145/1531326.1531378

    work page doi:10.1145/1531326.1531378 2009

  36. [38]

    In: IEEE In- ternational Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22- 29, 2017

    Litany, O., Remez, T., Rodolà, E., Bronstein, A.M., Bronstein, M.M.: Deep func- tional maps: Structured prediction for dense shape correspondence. In: IEEE In- ternational Conference on Computer Vision, ICCV 2017, Venice, Italy, October 22- 29, 2017. pp. 5660–5668. IEEE Computer Society, Washington, DC, USA (2017). https://doi.org/10.1109/ICCV.2017.603 , ...

    work page doi:10.1109/iccv.2017.603 2017

  37. [39]

    Loper, M., Mahmood, N., Romero, J., Pons-Moll, G., Black, M.J.: SMPL: A skinnedmulti-personlinearmodel.ACMTrans.Graphics(Proc.SIGGRAPHAsia) 34(6), 248:1–248:16 (Oct 2015)

    2015

  38. [40]

    Nguyen, A., Ben-Chen, M., Welnicka, K., Ye, Y., Guibas, L.J.: An optimization approach to improving collections of shape maps. Comput. Graph. Forum30(5), 1481–1491 (2011). https : / / doi . org / 10 . 1111 / J . 1467 - 8659 . 2011 . 02022 . X, https://doi.org/10.1111/j.1467-8659.2011.02022.x

    work page doi:10.1111/j.1467-8659.2011.02022.x 2011

  39. [41]

    ACM Trans

    Ovsjanikov, M., Ben-Chen, M., Solomon, J., Butscher, A., Guibas, L.: Functional maps: a flexible representation of maps between shapes. ACM Trans. Graph.31(4) (Jul 2012). https://doi.org/10.1145/2185520.2185526, https://doi.org/10. 1145/2185520.2185526

    work page doi:10.1145/2185520.2185526 2012

  40. [42]

    In: NIPS

    Pachauri, D., Kondor, R., Singh, V.: Solving the multi-way matching problem by permutation synchronization. In: NIPS. pp. 1860–1868 (2013)

    2013

  41. [43]

    In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019

    Park, J.J., Florence, P.R., Straub, J., Newcombe, R.A., Lovegrove, S.: Deepsdf: Learning continuous signed distance functions for shape representation. In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2019, Long Beach, CA, USA, June 16-20, 2019. pp. 165–174. Computer Vision Foundation / IEEE, New York, NY, USA (2019).https://doi.org/10...

    work page doi:10.1109/cvpr.2019.00025 2019

  42. [44]

    fast" -tune

    Pentapati, S.K., Phillips, G., Bovik, A.C.: Hierarchical neural surfaces for 3d mesh compression. In: 2025 Picture Coding Symposium (PCS). pp. 1–5 (2025).https: //doi.org/10.1109/PCS65673.2025.11417660

    work page doi:10.1109/pcs65673.2025.11417660 2025

  43. [45]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Pierson, E., Li, L., Dai, A., Ovsjanikov, M.: Diffumatch: Category-agnostic spec- tral diffusion priors for robust non-rigid shape matching. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 5745–5756. IEEE, Washington, DC, USA (October 2025)

    2025

  44. [46]

    arXiv:2507.23715 (2025)

    Pierson, E., Li, L., Dai, A., Ovsjanikov, M.: Diffumatch: Category-agnostic spectral diffusion priors for robust non-rigid shape matching. arXiv:2507.23715 (2025)

    work page arXiv 2025

  45. [47]

    ACM Trans

    Praun, E., Hoppe, H.: Spherical parametrization and remeshing. ACM Trans. Graph. 22(3), 340–349 (Jul 2003).https://doi.org/10.1145/882262.882274 , https://doi.org/10.1145/882262.882274

    work page doi:10.1145/882262.882274 2003

  46. [48]

    In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021

    Pumarola, A., Corona, E., Pons-Moll, G., Moreno-Noguer, F.: D-nerf: Neural ra- diance fields for dynamic scenes. In: IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021. pp. 10318–10327. Computer Vision Foundation / IEEE, Washington, DC, USA (2021). https: / / doi . org / 10 . 1109 / CVPR46437 . 2021 . 01018, h...

    2021

  47. [49]

    In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y

    Ranjan, A., Bolkart, T., Sanyal, S., Black, M.J.: Generating 3d faces using convo- lutional mesh autoencoders. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) Computer Vision - ECCV 2018 - 15th European Conference, Munich, 20 S.K. Pentapati et al. Germany, September 8-14, 2018, Proceedings, Part III. Lecture Notes in Com- puter Science, vo...

    work page doi:10.1007/978-3-030-01219-9_43 2018

  48. [50]

    Schmidt, P., Pieper, D., Kobbelt, L.: Surface maps via adaptive triangulations. Comput. Graph. Forum42(2), 103–117 (2023).https://doi.org/10.1111/CGF. 14747, https://doi.org/10.1111/cgf.14747

    work page doi:10.1111/cgf 2023

  49. [51]

    The Visual Computer 14(8-9), 429–444 (1998).https://doi.org/10.1007/s003710050143 , link.springer.com

    Shapiro, A., Tal, A.: Polygon realization for shape transformation. The Visual Computer 14(8-9), 429–444 (1998).https://doi.org/10.1007/s003710050143 , link.springer.com

    work page doi:10.1007/s003710050143 1998

  50. [52]

    Sheffer, A., Praun, E., Rose, K.: Mesh parameterization methods and their appli- cations. Found. Trends Comput. Graph. Vis.2(2), xx–yy (2006).https://doi. org/10.1561/0600000011, https://doi.org/10.1561/0600000011

    work page doi:10.1561/0600000011 2006

  51. [53]

    In: Lee, D.D., Sugiyama, M., von Luxburg, U., Guyon, I., Garnett, R

    Shen, Y., Huang, Q., Srebro, N., Sanghavi, S.: Normalized spectral map synchro- nization. In: Lee, D.D., Sugiyama, M., von Luxburg, U., Guyon, I., Garnett, R. (eds.) Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain. pp. 4925–4933 (2016),https://proc...

    2016

  52. [54]

    In: 2017 IEEE Conference on Computer Vi- sion and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017

    Slavcheva, M., Baust, M., Cremers, D., Ilic, S.: Killingfusion: Non-rigid 3d recon- struction without correspondences. In: 2017 IEEE Conference on Computer Vi- sion and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017. pp. 5474–5483. IEEE Computer Society, Washington, DC, USA (2017).https: //doi.org/10.1109/CVPR.2017.581, https://doi.or...

    work page doi:10.1109/cvpr.2017.581 2017

  53. [55]

    In: Proceedings of the Fifth Eurographics Symposium on Geometry Processing

    Sorkine, O., Alexa, M.: As-rigid-as-possible surface modeling. In: Proceedings of the Fifth Eurographics Symposium on Geometry Processing. pp. 109–116. SGP ’07, Eurographics Association, Goslar, DEU (2007)

    2007

  54. [56]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Sun, B., Groueix, T., Song, C., Huang, Q., Aigerman, N.: Tuttenet: Injective 3d deformations by composition of 2d mesh deformations. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 21378–21389. IEEE Computer Society, Washington, DC, USA (June 2024)

    2024

  55. [57]

    ACM Trans

    Sundararaman, R., Donati, N., Melzi, S., Corman, E., Ovsjanikov, M.: Deforma- tion recovery: Localized learning for detail-preserving deformations. ACM Trans. Graph. 43(6) (Nov 2024). https://doi.org/10.1145/3687968 , https://doi. org/10.1145/3687968

    work page doi:10.1145/3687968 2024

  56. [58]

    IEEE Trans

    Tam, G.K.L., Cheng, Z., Lai, Y., Langbein, F.C., Liu, Y., Marshall, A.D., Martin, R.R., Sun, X., Rosin, P.L.: Registration of 3d point clouds and meshes: A survey from rigid to nonrigid. IEEE Trans. Vis. Comput. Graph.19(7), 1199–1217 (2013). https://doi.org/10.1109/TVCG.2012.310 , https://doi.org/10.1109/TVCG. 2012.310

    work page doi:10.1109/tvcg.2012.310 2013

  57. [59]

    (eds.) Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, December 6-14, 2021, virtual

    Trappolini, G., Cosmo, L., Moschella, L., Marin, R., Melzi, S., Rodolà, E.: Shape registrationinthetimeoftransformers.In:Ranzato,M.,Beygelzimer,A.,Dauphin, Y.N., Liang, P., Vaughan, J.W. (eds.) Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, December 6-14, 2021, virtual. ...

    2021

  58. [60]

    Frozen in time: A joint video and image encoder for end-to-end retrieval

    Tretschk, E., Tewari, A., Golyanik, V., Zollhöfer, M., Lassner, C., Theobalt, C.: Non-rigid neural radiance fields: Reconstruction and novel view synthesis of a dy- namic scene from monocular video. In: 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021, Montreal, QC, Canada, October 10-17, 2021. pp. GenSP: Consistent Spherical Parameter...

    work page doi:10.1109/iccv48922.2021.01272 2021

  59. [61]

    Wang, C., Liu, Z., Liu, L.: As-rigid-as-possible spherical parametrization. Graph. Models 76(5), 457–467 (Sep 2014).https://doi.org/10.1016/j.gmod.2014.03. 016, https://doi.org/10.1016/j.gmod.2014.03.016

    work page doi:10.1016/j.gmod.2014.03 2014

  60. [62]

    Williamson, R., Mitra, N.J.: Neural geometry processing via spherical neural sur- faces. Comput. Graph. Forum44(2), xx–yy (2025).https://doi.org/10.1111/ CGF.70021, https://doi.org/10.1111/cgf.70021

    work page doi:10.1111/cgf.70021 2025

  61. [63]

    In: The Twelfth International Conference on Learning Representations, ICLR 2024, Vi- enna, Austria, May 7-11, 2024

    Yang, H., Huang, X., Sun, B., Bajaj, C.L., Huang, Q.: Gencorres: Consistent shape matching via coupled implicit-explicit shape generative models. In: The Twelfth International Conference on Learning Representations, ICLR 2024, Vi- enna, Austria, May 7-11, 2024. pp. xx–yy. OpenReview.net, Vienna, Austria (2024), https://openreview.net/forum?id=dGH4kHFKFj

    2024

  62. [64]

    ACM Trans

    Yang,H.,Sun,B.,Chen,L.,Pavel,A.,Huang,Q.:Geolatent:Ageometricapproach to latent space design for deformable shape generators. ACM Trans. Graph.42(6) (Dec 2023). https://doi.org/10.1145/3618371 , https://doi.org/10.1145/ 3618371

    work page doi:10.1145/3618371 2023

  63. [65]

    ACM Trans

    Yang, Y., Yang, H., Nakayama, K., Huang, X., Guibas, L.J., Huang, Q.: Gen- analysis: Joint shape analysis by learning man-made shape generators with de- formation regularizations. ACM Trans. Graph.44(4), 61:1–61:19 (2025).https: //doi.org/10.1145/3731164, https://doi.org/10.1145/3731164

    work page doi:10.1145/3731164 2025

  64. [66]

    Computer Graphics Forum33(5), 257–267 (2014).https://doi.org/10

    Yoshiyasu, Y., Ma, L., Yoshida, E., Gu, X.: As-conformal-as-possible surface regis- tration. Computer Graphics Forum33(5), 257–267 (2014).https://doi.org/10. 1111/cgf.12451

    2014

  65. [67]

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

    You, Y., Lou, Y., Li, C., Cheng, Z., Li, L., Ma, L., Lu, C., Wang, W.: Keypointnet: A large-scale 3d keypoint dataset aggregated from numerous human annotations. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 13647–13656 (2020)

    2020

  66. [68]

    In: Vedaldi, A., Bischof, H., Brox, T., Frahm, J

    Zhou, K., Bhatnagar, B.L., Pons-Moll, G.: Unsupervised shape and pose dis- entanglement for 3d meshes. In: Vedaldi, A., Bischof, H., Brox, T., Frahm, J. (eds.) Computer Vision - ECCV 2020 - 16th European Conference, Glasgow, UK, August 23-28, 2020, Proceedings, Part XXII. Lecture Notes in Computer Science, vol. 12367, pp. 341–357. Springer, New York, NY, ...

    work page doi:10.1007/978-3-030-58542-6_21 2020

  67. [69]

    In: Proceedings of the IEEE International Conference on Computer Vision

    Zhou, X., Zhu, M., Daniilidis, K.: Multi-image matching via fast alternating min- imization. In: Proceedings of the IEEE International Conference on Computer Vision. pp. 4032–4040 (2015)

    2015

  68. [70]

    In: Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., Lin, H

    Zhou, Y., Wu, C., Li, Z., Cao, C., Ye, Y., Saragih, J.M., Li, H., Sheikh, Y.: Fully convolutional mesh autoencoder using efficient spatially varying kernels. In: Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., Lin, H. (eds.) Advances in Neural Information Processing Systems 33: Annual Con- ference on Neural Information Processing Systems 2020, NeurI...

    2020

  69. [71]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025

    Zhuravlev, A., Lähner, Z., Golyanik, V.: Denoising functional maps: Diffusion models for shape correspondence. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. pp. 26899–26909. Computer Vision Foundation / IEEE, Washington, DC, USA (2025). https : / / doi . org / 10 . 1109 / CVPR52734 . 2...

    2025

  70. [72]

    In: 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017

    Zuffi, S., Kanazawa, A., Jacobs, D.W., Black, M.J.: 3d menagerie: Modeling the 3d shape and pose of animals. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017. pp. 5524–

    2017

  71. [73]

    6 This section presents a matrix reformulation of Eq

    IEEE Computer Society, New York, NY, USA (2017).https://doi.org/ 10.1109/CVPR.2017.586, https://doi.org/10.1109/CVPR.2017.586 GenSP: Consistent Spherical Parameterizations 23 A Derivation of Eq. 6 This section presents a matrix reformulation of Eq. 6, which is dT Lϕ(z)d = min si,ci lX i=1 X j∈Ni ∥Ai(pϕ i (z) − pϕ j (z)) − (di − dj)∥2 + ηs2 i (13) where Ai...

    work page doi:10.1109/cvpr.2017.586 2017