Pith. sign in

REVIEW 5 major objections 8 minor 58 references

FUSE: A Flow-based Mapping Between Shapes

T0 review · 5 major / 8 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read FUSE represents the map between any two 3D shapes as the composition of two independently trained flow-matching models through a shared Gaussian anchor, producing bijective correspondences without per-pair optimization.

desk verdict FUSE is a genuinely new zero-shot representation for shape maps built on per-shape flow matching, with broad and mostly solid empirical results; the anchor-alignment premise is the real soft spot, but the empirics carry it. read the letter →

arxiv 2511.13431 v2 pith:EMA4LCSE submitted 2025-11-17 cs.CV

Lorenzo Olearo , Giulio Viganò , Daniele Baieri , Filippo Maggioli , Simone Melzi This is my paper · ORCID
classification cs.CV
keywords shapematchingflowcorrespondencecross-representationgeodesicembeddingsinvertibleflowszero-shotprobabilisticalignment
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 introduces FUSE, a neural representation for correspondences between 3D shapes that treats each shape's feature distribution as something to be transported to a common Gaussian anchor by an invertible flow. Once each shape has its own flow, matching any pair becomes the composition of the source's inverse flow and the target's forward flow — a zero-shot operation that requires no further training or optimization. The authors argue that this probabilistic alignment improves correspondence accuracy whenever the underlying features are not already perfectly aligned, and they demonstrate the construction across meshes, point clouds, signed distance fields, and volumetric data. The significance is that shape matching becomes a one-time-per-shape investment that then spans representations and deformations, with bijectivity and coverage built in.

What carries the argument

The central object is the flow-matching model ψ: [0,1]×R^d→R^d, an invertible diffeomorphism trained via Conditional Flow Matching to transport samples from a fixed Gaussian anchor p0 = N(0,1)^d to a shape's pointwise embedding distribution pℓ. The map between shapes is the composition Φ12 = ψ₂ ∘ (ψ₁)⁻¹: the inverse flow pulls source embeddings back to the shared Gaussian, and the forward flow pushes them into the target's embedding space. The shared Gaussian is what makes the construction representation-agnostic and zero-shot, since any two flows from the same anchor can be composed without additional optimization. The geodesic feature embedding (distances to a handful of landmarks) is the

What would settle it

For a benchmark with dense ground-truth correspondences, compare the Euclidean distances in the Gaussian anchor between ψ₁⁻¹(E1(x)) and ψ₂⁻¹(E2(y)) for true corresponding pairs (x,y) against the same distances for random point pairs; the central claim collapses if these two distance distributions are statistically indistinguishable, since the anchor then carries no correspondence information.

Watch

Extended reading notes

Core claim

The central claim is that, given two trained flows ψ₁ and ψ₂ from a shared Gaussian anchor to the embedding distributions of S₁ and S₂, the composition Φ12 = ψ₂ ∘ (ψ₁)⁻¹ is an accurate, bijective, high-coverage pointwise map from S₁ to S₂. Because each flow is trained independently on its own shape with no cross-shape constraint, the method is a purely generative representation: corresponding points end up near each other in the anchor because the flow-matching objective makes distributional alignment coincide with semantic alignment. The paper argues that this holds for any embedding that encodes shape semantics, and that the flow composition acts as a probabilistic alignment mechanism — im

Load-bearing premise

The load-bearing premise is that independently trained flows map semantically corresponding points to nearby locations in the shared Gaussian anchor; if the flows from different shapes do not cluster corresponding points in the anchor, the composition Φ12 degenerates to a random bijection between distributions and gains nothing over direct nearest-neighbor search.

Editorial extensions

If this is right

  • Once each shape has its flow, any pair of shapes becomes matchable by composing flows, with no per-pair optimization or learned correspondence priors.
  • The representation is modality-agnostic: because only surface sampling and embedding computation are required, the same flows support mesh-to-mesh, point-cloud-to-mesh, SDF-to-mesh, and volumetric matching.
  • Bijectivity and high coverage are inherited from the invertibility of flows, improving on functional-map and optimal-transport baselines in coverage while achieving competitive accuracy.
  • The flow acts as a probabilistic aligner: it consistently improves accuracy and distributional similarity for moderately aligned embeddings (e.g., geodesic features), and it can initialize refinement methods like ZoomOut and Neural Adjoint Maps to state-of-the-art results.
  • Beyond correspondence, the construction serves as a backbone for UV parametrization, human-scan reposing, and volume matching.

Reading between the lines

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

  • The same anchored-flow construction suggests a universal indexing scheme for shape collections: if every shape maps to the shared Gaussian, then nearest-neighbor search in the anchor space could support retrieval, interpolation, and shape analogy without ad hoc matching — a direction the paper leaves implicit.
  • Because the benefit is tied to distributional misalignment, a testable extension is to characterize the regime of feature alignment in which the flow bridge is net positive, and to intentionally use features that are order-invariant but distributionally misaligned so the flow does the alignment work.
  • If the core locality conjecture holds, flow-matching over feature distributions is a general alignment primitive that could port to other domains where semantically linked points live in different distributions across instances (e.g., audio or text embedding alignment), not just 3D geometry.
  • The anchored Gaussian also makes the representation transitive: composing ψ₃∘(ψ₁)⁻¹ can be viewed as a chain through intermediate shapes, suggesting a compositional structure that functional maps lack, though the paper does not explore it.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 8 minor

Summary. FUSE proposes a neural representation for maps between 3D shapes based on flow matching. For each shape, a flow model is trained independently to transport the shape's pointwise feature embedding distribution to a shared Gaussian anchor. The map between two shapes is then defined as the composition of the source inverse flow and the target forward flow (Eq. 2). The method claims zero-shot, representation-agnostic matching across meshes, point clouds, SDFs, and volumes, and reports competitive accuracy on FAUST, SMAL, SHREC20, Kinect, and additional tasks including UV mapping, human scan registration, and volumetric correspondence. The central premise is that independently trained flows map semantically corresponding points to nearby locations in the Gaussian anchor, enabling the composed map to align embeddings without any cross-shape supervision.

Significance. If the central premise holds, FUSE offers an elegant and novel probabilistic view of shape maps: a per-shape flow representation that supports zero-shot composition for arbitrary pairs and representations. The idea of using flow matching over feature embeddings rather than raw coordinates is a meaningful extension of GeomDist, and the empirical breadth—spanning several benchmarks and modalities—is commendable. The method is simple, computationally lightweight per shape, and yields high coverage by construction if the flows are truly invertible. However, the paper's main claim rests on an unproven alignment property of independent flow solutions, and at least one experimental result (FMNet embeddings in Table 1) directly contradicts the stated universality. The paper would benefit from a precise theoretical or empirical characterization of when the anchor-alignment property holds, as well as a more careful framing of its claims.

major comments (5)
  1. [§4.2, Eq. (2)] The load-bearing assumption that independently trained flows map corresponding points to nearby anchor locations is not established. The CFM objective (Eq. 1) only constrains the per-shape pushforward p_0 to p_l; it does not constrain the joint law of (ψ_1^{-1}(f_1(x)), ψ_2^{-1}(f_2(y))) for corresponding x,y. FM solutions are non-unique, and the stochastic pairing in training can produce arbitrary measure-preserving rearrangements. The sentence after Eq. (3) merely asserts the property. Please provide either a theoretical condition (e.g., uniqueness of the optimal transport map) or direct empirical evidence: for known correspondences, measure the distance between latent codes, compare the composition against a random bijection, and test stability across training seeds. Without this, Eq. (2) could in principle be equivalent to random matching.
  2. [§4.3, Table 1] The claim that FUSE is a general alignment mechanism is contradicted by the FMNet row: FUSE increases the correspondence error from 0.000062 (KNN) to 0.001158, a degradation of nearly 20x. The text acknowledges this ('may even slightly corrupt them'), but such a caveat undermines the abstract's claim of a universal representation. The authors should either restrict the claim to embeddings that are not already well aligned, or provide a quantitative criterion—e.g., distributional similarity or a learned classifier—that predicts when FUSE will help. As is, the method is not universally beneficial across embeddings, which is a central advertised property.
  3. [§4.1, §4.4] The paper states that flow matching 'guarantees that the inverse flow deterministically maps any shape back to the exact Gaussian.' In practice, the learned velocity field is a neural network and the ODE integration is numerical; the inverse map is only approximate. Table 2 measures KL/JS divergence after inversion for geodesic embeddings, but this is a single embedding and the analysis does not quantify how inversion error propagates into the composition Φ_{12}. Please report inversion accuracy across embeddings and shapes, and discuss the effect of integration tolerance on the final correspondence accuracy. The distinction from GeomDist's approximate inversion is quantitative, not qualitative, and needs a controlled comparison.
  4. [§5.1, metrics] The 'Coverage' metric is never formally defined, yet it is used to argue that FUSE achieves better coverage than baselines. If coverage is simply the size of the image of the discrete map, then any surjective map (or a map defined as ψ_2∘ψ_1^{-1} applied to a dense enough sample) trivially achieves high coverage, making the comparison with functional-map methods that produce soft or low-rank maps unfair. Please provide the exact definition, how it is computed from the map matrix, and what property it is intended to measure. The statement that FUSE achieves better coverage 'due to the nature of the flows' is circular unless coverage is defined independently of the map's construction.
  5. [§5.1, landmark selection] The geodesic embedding relies on a small set of landmarks (5 for FAUST/SMAL, 6 for SHREC20) selected 'at the protrusions' or 'common landmarks.' The procedure for choosing these landmarks is not specified: Are they placed manually, automatically, or via a semantic prior? If landmarks are chosen to correspond across shapes, then the embedding itself already encodes correspondence information, and the 'no learned priors' claim is weakened. The sensitivity of FUSE to the number and choice of landmarks is not analyzed, yet the landmark embedding is the primary unsupervised descriptor used in the main experiments. Please clarify the selection protocol and provide an ablation over landmark sets.
minor comments (8)
  1. [Related Work] Typo: 'architechtures' should be 'architectures'.
  2. [§5.2, Inter-representation] Typo: 'gedesics' should be 'geodesics'.
  3. [§5.1, Datasets] The Kinect dataset description says 'intoduced' instead of 'introduced'.
  4. [Table 5] 'V olumetric' has a stray space.
  5. [§5.1, SHREC20] The phrase 'we consider all the pairs of the 5 test set' is unclear. Please specify the test set size and the exact pairing protocol.
  6. [§5.1, Baselines] The KNN-IN-GAUSS baseline is not defined precisely. Does it invert the flow using the same ODE solver as FUSE? If not, the comparison may conflate inversion error with alignment quality.
  7. [General] No code or supplement link is provided. Given the number of implementation details (flow architecture, training hyperparameters, ODE solver settings), the paper would benefit from releasing code or a detailed supplementary to ensure reproducibility.
  8. [§4.4] The reference to [16] as 'The Fourth Blogpost Track at ICLR 2025' is informal; consider citing the formal equivalence paper (e.g., Lipman et al. or follow-up work).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FUSE's map is a composition of independently trained flows with no parameter fitted to correspondence evaluation.

full rationale

The paper's core construction is Eq. (2), composing two independently trained flow-matching models. Each flow is optimized only against its own shape's embedding distribution via the CFM objective in Eq. (1), with no supervision from correspondences, target maps, or any benchmark metric. The resulting map and nearest-neighbor search in Eq. (3) are then evaluated, not fitted. The claim that semantic correspondences tend to share latent coordinates is an empirical hypothesis about the behavior of independently trained flows, not a step that reduces to its inputs by definition. The paper's self-citations (e.g., [28], [52]) appear as baselines or refinement components, not as load-bearing justification for the central construction. The concern that the anchor-alignment premise is unproven is a correctness/validity issue, not circularity, and does not meet the bar of exhibiting a specific reduction of a prediction to fitted inputs.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or mathematical entities; it uses existing flow-matching machinery and standard geodesic embeddings. The load-bearing assumptions are the accuracy of the embedding distributions and the empirical alignment property of independently trained flows.

free parameters (2)
  • Landmark set size and selection (geodesic embedding) = 5 landmarks for humans, 6 for animals
    The geodesic feature embedding is computed as distances to a hand-picked set of protrusion landmarks (fingertips, nose, etc.). The number and choice of landmarks are not learned or derived, and they inject semantic correspondence information; different landmarks would change the results.
  • Flow-matching network architecture and training hyperparameters
    The flow network architecture (depth, width, activation), optimizer, learning rate, and number of training iterations are not reported, so the practitioner must choose them ad hoc; these choices likely affect the quality of the learned flows and thus the final maps.
assumptions (4)
  • standard math Conditional Flow Matching (CFM) provides a velocity field whose integral flow maps a Gaussian to the given data distribution (Lipman et al. 2022).
    The method relies on the CFM framework to learn the per-shape flows ψₓ.
  • domain assumption The pointwise embedding distributions p₁ and p₂ are smooth, continuous probability distributions on Rᵈ so that flow matching can be applied and the flows are diffeomorphisms.
    FUSE treats the discrete embedding matrices as distributions; if the sampled points are not sufficiently dense or the features are discontinuous, the learned flow may be ill-defined. Section 4.1 assumes sampling from the surface.
  • domain assumption Geodesic distance computations are accurate on all considered representations (heat method on point clouds, Dijkstra on meshes and voxel grids).
    The geodesic embedding used as the main feature relies on single-to-all geodesic distances; errors in these distances on point clouds/SDFs propagate to the flow and the final correspondence (Section 5.1 notes degraded performance on point clouds due to unreliable geodesics).
  • ad hoc to paper Points with similar embeddings from different shapes map to nearby regions in the shared Gaussian anchor after independently trained flows, so composition ψ₂∘ψ₁⁻¹ yields semantically correct correspondences.
    This is the paper's core probabilistic-alignment assumption (Section 4.2, green box). The flows are trained independently per shape with no cross-shape consistency constraint; whether similar embeddings land in the same anchor region is not guaranteed in theory, only empirically observed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FUSE: A Flow-based Mapping Between Shapes." pith.science (2026). https://pith.science/paper/EMA4LCSE

@misc{pith2026251113431,
  author       = {Pith},
  title        = {Pith review of: FUSE: A Flow-based Mapping Between Shapes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EMA4LCSE}},
  note         = {Machine review of arXiv:2511.13431}
}
read the original abstract

We introduce a novel neural representation for maps between 3D shapes based on flow-matching models, which is computationally efficient and supports cross-representation shape matching without large-scale training or data-driven procedures. 3D shapes are represented as the probability distribution induced by a continuous and invertible flow mapping from a fixed anchor distribution. Given a source and a target shape, the composition of the inverse flow (source to anchor) with the forward flow (anchor to target), we map points between the two surfaces. By encoding the shapes with a pointwise task-tailored embedding, this construction provides an invertible and modality-agnostic representation of maps between shapes across point clouds, meshes, signed distance fields (SDFs), and volumetric data. The resulting representation consistently achieves high coverage and accuracy across diverse benchmarks and challenging settings in shape matching. Beyond shape matching, our framework shows promising results in other tasks, including UV mapping and registration of raw point cloud scans of human bodies.

Figures

Figures reproduced from arXiv: 2511.13431 by the authors.

Figure 1
Figure 1. Our representation establishes a map between shapes [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Visual example of the proposed pipeline. For visualization purpose, we select a pair for which we can solve for the correspondence [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Effect of FUSE visualized on the first three coordinates [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Correspondences obtained by FUSE with different em [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Qualitative evaluation: We report a pair of shapes with visualization of the p2p as transfered colormap and in terms of the [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Coordinate transfer among different 3D representations [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 8
Figure 8. Figure 8: Illustration of our parametrization via texture mapping. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 2 linked inside Pith

  1. [1]

    Kim, Siddhartha Chaudhuri, Jun Saito, and Thibault Groueix

    Noam Aigerman, Kunal Gupta, Vladimir G. Kim, Siddhartha Chaudhuri, Jun Saito, and Thibault Groueix. Neural jacobian fields: learning intrinsic mappings of arbitrary meshes.ACM Trans. Graph., 41(4), 2022. 2

    2022

  2. [2]

    The wave kernel signature: A quantum mechanical approach to shape analysis

    Mathieu Aubry, Ulrich Schlickewei, and Daniel Cremers. The wave kernel signature: A quantum mechanical approach to shape analysis. InComputer Vision Workshops (ICCV Workshops), 2011 IEEE International Conference on, pages 1626–1633. IEEE, 2011. 2, 3, 4, 5, 6

    2011

  3. [3]

    Implicit-arap: Efficient handle-guided deformation of high-resolution meshes and neural fields via local patch meshing, 2024

    Daniele Baieri, Filippo Maggioli, Zorah L ¨ahner, Simone Melzi, and Emanuele Rodol `a. Implicit-arap: Efficient handle-guided deformation of high-resolution meshes and neural fields via local patch meshing, 2024. 2, 7

    2024

  4. [4]

    Hybrid functional maps for crease-aware non- isometric shape matching, 2024

    Lennart Bastian, Yizheng Xie, Nassir Navab, and Zorah L¨ahner. Hybrid functional maps for crease-aware non- isometric shape matching, 2024. 2

    2024

  5. [5]

    Behave: Dataset and method for tracking human object inter- actions

    Bharat Lal Bhatnagar, Xianghui Xie, Ilya Petrov, Cristian Sminchisescu, Christian Theobalt, and Gerard Pons-Moll. Behave: Dataset and method for tracking human object inter- actions. InIEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2022. 6

    2022

  6. [6]

    Faust: Dataset and evaluation for 3d mesh registration

    Federica Bogo, Javier Romero, Matthew Loper, and Michael J Black. Faust: Dataset and evaluation for 3d mesh registration. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 3794– 3801, 2014. 6

    2014

  7. [7]

    Learning class-specific descriptors for deformable shapes us- ing localized spectral convolutional networks.Computer Graphics Forum, 34(5):13–23, 2015

    Davide Boscaini, Jonathan Masci, Simone Melzi, Michael M Bronstein, Umberto Castellani, and Pierre Vandergheynst. Learning class-specific descriptors for deformable shapes us- ing localized spectral convolutional networks.Computer Graphics Forum, 34(5):13–23, 2015. 2

    2015

  8. [8]

    Unsu- pervised learning of robust spectral shape matching.ACM Trans

    Dongliang Cao, Paul Roetzer, and Florian Bernard. Unsu- pervised learning of robust spectral shape matching.ACM Trans. Graph., 42(4), 2023. 2, 4, 6

    2023

Show all 58 references
  1. [9]

    Spectral meets spatial: Harmonising 3d shape matching and interpolation, 2024

    Dongliang Cao, Marvin Eisenberger, Nafie El Amrani, Daniel Cremers, and Florian Bernard. Spectral meets spatial: Harmonising 3d shape matching and interpolation, 2024. 2

    2024

  2. [10]

    The heat method for distance computation.Commun

    Keenan Crane, Clarisse Weischedel, and Max Wardetzky. The heat method for distance computation.Commun. ACM, 60(11):90–99, 2017. 4, 7

    2017

  3. [11]

    Sinkhorn distances: lightspeed computation of optimal transport

    Marco Cuturi. Sinkhorn distances: lightspeed computation of optimal transport. InProceedings of the 27th International Conference on Neural Information Processing Systems - Vol- ume 2, page 2292–2300, Red Hook, NY , USA, 2013. Curran Associates Inc. 2, 5, 6

    2013

  4. [12]

    Deep orientation-aware functional maps: Tackling symme- try issues in shape matching.CVPR, 2022

    Nicolas Donati, Etienne Corman, and Maks Ovsjanikov. Deep orientation-aware functional maps: Tackling symme- try issues in shape matching.CVPR, 2022. 2, 4

    2022

  5. [13]

    Niladri Shekhar Dutt, Sanjeev Muralikrishnan, and Niloy J. Mitra. Diffusion 3d features (diff3f): Decorating untextured shapes with distilled semantic features. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 4494–4504, 2024. 3

    2024

  6. [14]

    Dyke, Yu-Kun Lai, Paul L

    Roberto M. Dyke, Yu-Kun Lai, Paul L. Rosin, Stefano Zap- pal`a, Seana Dykes, Daoliang Guo, Kun Li, Riccardo Marin, Simone Melzi, and Jingyu Yang. Shrec20: Shape corre- spondence with non-isometric deformations.Computers & Graphics, 92:28–43, 2020. 6

    2020

  7. [15]

    Neuromorph: Unsupervised shape interpola- tion and correspondence in one go, 2021

    Marvin Eisenberger, David Novotny, Gael Kerchenbaum, Patrick Labatut, Natalia Neverova, Daniel Cremers, and An- drea Vedaldi. Neuromorph: Unsupervised shape interpola- tion and correspondence in one go, 2021. 1

    2021

  8. [16]

    Diffusion models and gaussian flow matching: Two sides of the same coin

    Ruiqi Gao, Emiel Hoogeboom, Jonathan Heek, Valentin De Bortoli, Kevin Patrick Murphy, and Tim Salimans. Diffusion models and gaussian flow matching: Two sides of the same coin. InThe Fourth Blogpost Track at ICLR 2025, 2025. 5

    2025

  9. [17]

    Non-rigid shape registration via deep functional maps prior

    Puhua Jiang, Mingze Sun, and Ruqi Huang. Non-rigid shape registration via deep functional maps prior. InThirty- seventh Conference on Neural Information Processing Sys- tems, 2023. 2

    2023

  10. [18]

    Blended intrinsic maps

    Vladimir G Kim, Yaron Lipman, and Thomas Funkhouser. Blended intrinsic maps. InACM Transactions on Graphics (TOG), page 79. ACM, 2011. 6

    2011

  11. [19]

    Intrinsic neural fields: Learning functions on manifolds

    Lukas Koestler, Daniel Grittner, Michael Moeller, Daniel Cremers, and Zorah L¨ahner. Intrinsic neural fields: Learning functions on manifolds. InComputer Vision – ECCV 2022, pages 622–639, Cham, 2022. Springer Nature Switzerland. 2

    2022

  12. [20]

    Integrating efficient optimal transport and functional maps for unsupervised shape correspondence learning

    Tung Le, Khai Nguyen, Shanlin Sun, Nhat Ho, and Xiaohui Xie. Integrating efficient optimal transport and functional maps for unsupervised shape correspondence learning. In 2024 IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), pages 23188–23198, 2024. 2

    2024

  13. [21]

    Non-rigid point cloud registra- tion with neural deformation pyramid

    Yang Li and Tatsuya Harada. Non-rigid point cloud registra- tion with neural deformation pyramid. InProceedings of the 36th International Conference on Neural Information Pro- cessing Systems, Red Hook, NY , USA, 2022. Curran Asso- ciates Inc. 2, 5, 6

    2022

  14. [22]

    Flow matching for generative mod- eling.arXiv preprint arXiv:2210.02747, 2022

    Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maximil- ian Nickel, and Matt Le. Flow matching for generative mod- eling.arXiv preprint arXiv:2210.02747, 2022. 2, 3

    2022 arXiv

  15. [23]

    Flow matching guide and code.arXiv preprint arXiv:2412.06264, 2024

    Yaron Lipman, Marton Havasi, Peter Holderrieth, Neta Shaul, Matt Le, Brian Karrer, Ricky TQ Chen, David Lopez- Paz, Heli Ben-Hamu, and Itai Gat. Flow matching guide and code.arXiv preprint arXiv:2412.06264, 2024. 3

    2024 arXiv

  16. [24]

    Zero-1-to- 3: Zero-shot one image to 3d object

    Ruoshi Liu, Rundi Wu, Basile Van Hoorick, Pavel Tok- makov, Sergey Zakharov, and Carl V ondrick. Zero-1-to- 3: Zero-shot one image to 3d object. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 9298–9309, 2023. 2

    2023

  17. [25]

    Editing conditional radiance fields

    Steven Liu, Xiuming Zhang, Zhoutong Zhang, Richard Zhang, Jun-Yan Zhu, and Bryan Russell. Editing conditional radiance fields. InProceedings of the International Confer- ence on Computer Vision (ICCV), 2021. 2

    2021

  18. [26]

    Semantic correspondence as an optimal transport problem

    Yanbin Liu, Linchao Zhu, Makoto Yamada, and Yi Yang. Semantic correspondence as an optimal transport problem. 9 InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4463–4472, 2020. 2

    2020

  19. [27]

    Rematching: Low-resolution representations for scalable shape correspondence

    Filippo Maggioli, Daniele Baieri, Emanuele Rodol `a, and Si- mone Melzi. Rematching: Low-resolution representations for scalable shape correspondence. InEuropean Conference on Computer Vision, pages 183–200. Springer, 2025. 2

    2025

  20. [28]

    V olu- metric functional maps, 2025

    Filippo Maggioli, Simone Melzi, and Marco Livesu. V olu- metric functional maps, 2025. 8

    2025

  21. [29]

    Smooth non-rigid shape matching via effective dirichlet energy optimization

    Robin Magnet, Jing Ren, Olga Sorkine-Hornung, and Maks Ovsjanikov. Smooth non-rigid shape matching via effective dirichlet energy optimization. In2022 International Confer- ence on 3D Vision (3DV). IEEE, 2022. 6

    2022

  22. [30]

    CMH: Coor- dinates Manifold Harmonics for Functional Remeshing

    Riccardo Marin, Simone Melzi, Pietro Musoni, Filippo Bar- don, Marco Tarini, and Umberto Castellani. CMH: Coor- dinates Manifold Harmonics for Functional Remeshing. In Eurographics Workshop on 3D Object Retrieval. The Euro- graphics Association, 2019. 2

    2019

  23. [31]

    Correspondence learning via linearly- invariant embedding

    Riccardo Marin, Marie-Julie Rakotosaona, Simone Melzi, and Maks Ovsjanikov. Correspondence learning via linearly- invariant embedding. InProceedings of the 34th Interna- tional Conference on Neural Information Processing Sys- tems, Red Hook, NY , USA, 2020. Curran Associates Inc. 2

    2020

  24. [32]

    Discrete time evolution process descriptor for shape analysis and matching.ACM Transactions on Graphics (TOG), 37(1):4:1–4:18, 2018

    Simone Melzi, Maks Ovsjanikov, Giorgio Roffo, Marco Cristani, and Umberto Castellani. Discrete time evolution process descriptor for shape analysis and matching.ACM Transactions on Graphics (TOG), 37(1):4:1–4:18, 2018. 2

    2018

  25. [33]

    Zoomout: Spectral upsampling for efficient shape correspondence

    Simone Melzi, Jing Ren, Emanuele Rodol `a, Abhishek Sharma, Peter Wonka, and Maks Ovsjanikov. Zoomout: Spectral upsampling for efficient shape correspondence. ACM Transactions on Graphics (TOG), 38(6):155:1–155:14,

  26. [34]

    Occupancy networks: Learning 3d reconstruction in function space

    Lars Mescheder, Michael Oechsle, Michael Niemeyer, Se- bastian Nowozin, and Andreas Geiger. Occupancy networks: Learning 3d reconstruction in function space. InProceed- ings IEEE Conf. on Computer Vision and Pattern Recogni- tion (CVPR), 2019. 2

    2019

  27. [35]

    Srinivasan, Matthew Tancik, Jonathan T

    Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, and Ren Ng. Nerf: Representing scenes as neural radiance fields for view syn- thesis. InECCV, 2020. 1, 2

    2020

  28. [36]

    Fast computation of weighted distance functions and geodesics on implicit hyper- surfaces.Journal of Computational Physics, 173(2):730– 764, 2001

    Facundo M ´emoli and Guillermo Sapiro. Fast computation of weighted distance functions and geodesics on implicit hyper- surfaces.Journal of Computational Physics, 173(2):730– 764, 2001. 4

    2001

  29. [37]

    Neural im- plicit surface evolution

    Tiago Novello, Vin ´ıcius da Silva, Guilherme Schardong, Luiz Schirmer, H ´elio Lopes, and Luiz Velho. Neural im- plicit surface evolution. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 14279–14289, 2023. 2

    2023

  30. [38]

    Functional maps: a flexible representation of maps between shapes.ACM Transactions on Graphics (TOG), 31(4):30:1–30:11, 2012

    Maks Ovsjanikov, Mirela Ben-Chen, Justin Solomon, Adrian Butscher, and Leonidas Guibas. Functional maps: a flexible representation of maps between shapes.ACM Transactions on Graphics (TOG), 31(4):30:1–30:11, 2012. 2, 3, 4, 5, 6

    2012

  31. [39]

    G. Pai, J. Ren, S. Melzi, P. Wonka, and M. Ovsjanikov. Fast sinkhorn filters: Using matrix scaling for non-rigid shape correspondence with functional maps. In2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 384–393, Los Alamitos, CA, USA, 2021....

    2021

  32. [40]

    Brunton, and J

    Shaowu Pan, Steven L. Brunton, and J. Nathan Kutz. Neu- ral implicit flow: a mesh-agnostic dimensionality reduction paradigm of spatio-temporal data.J. Mach. Learn. Res., 24 (1), 2023. 2

    2023

  33. [41]

    Deepsdf: Learning con- tinuous signed distance functions for shape representation

    Jeong Joon Park, Peter Florence, Julian Straub, Richard Newcombe, and Steven Lovegrove. Deepsdf: Learning con- tinuous signed distance functions for shape representation. InProceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 165–174, 2019. 1, 2

    2019

  34. [42]

    Georgios Pavlakos, Vasileios Choutas, Nima Ghorbani, Timo Bolkart, Ahmed A. A. Osman, Dimitrios Tzionas, and Michael J. Black. Expressive body capture: 3D hands, face, and body from a single image. InProceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), pag...

    2019

  35. [43]

    Barron, and Ben Milden- hall

    Ben Poole, Ajay Jain, Jonathan T. Barron, and Ben Milden- hall. Dreamfusion: Text-to-3d using 2d diffusion. InThe Eleventh International Conference on Learning Representa- tions, 2023. 2

    2023

  36. [44]

    At- tention And Positional Encoding Are (Almost) All You Need For Shape Matching.Computer Graphics Forum, 2023

    Alessandro Raganato, Gabriella Pasi, and Simone Melzi. At- tention And Positional Encoding Are (Almost) All You Need For Shape Matching.Computer Graphics Forum, 2023. 2

    2023

  37. [45]

    Bending graphs: Hierarchical shape matching using gated optimal transport,

    Mahdi Saleh, Shun-Cheng Wu, Luca Cosmo, Nassir Navab, Benjamin Busam, and Federico Tombari. Bending graphs: Hierarchical shape matching using gated optimal transport,

  38. [46]

    Diffusionnet: Discretization agnostic learning on surfaces.ACM Trans

    Nicholas Sharp, Souhaib Attaiki, Keenan Crane, and Maks Ovsjanikov. Diffusionnet: Discretization agnostic learning on surfaces.ACM Trans. Graph., 41(3), 2022. 4

    2022

  39. [47]

    3d pose transfer with correspondence learning and mesh refinement, 2021

    Chaoyue Song, Jiacheng Wei, Ruibo Li, Fayao Liu, and Gu- osheng Lin. 3d pose transfer with correspondence learning and mesh refinement, 2021. 1

    2021

  40. [48]

    Denois- ing diffusion implicit models, 2022

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denois- ing diffusion implicit models, 2022. 2

    2022

  41. [49]

    A con- cise and provably informative multi-scale signature based on heat diffusion.Computer graphics forum, 28(5):1383–1392,

    Jian Sun, Maks Ovsjanikov, and Leonidas Guibas. A con- cise and provably informative multi-scale signature based on heat diffusion.Computer graphics forum, 28(5):1383–1392,

  42. [50]

    Unique signatures of histograms for local surface descrip- tion

    Federico Tombari, Samuele Salti, and Luigi Di Stefano. Unique signatures of histograms for local surface descrip- tion. InProc. ECCV, pages 356–369. Springer, 2010. 2, 3

    2010

  43. [51]

    Shape registration in the time of transformers.Advances in Neural Information Processing Systems, 34:5731–5744, 2021

    Giovanni Trappolini, Luca Cosmo, Luca Moschella, Ric- cardo Marin, Simone Melzi, and Emanuele Rodol `a. Shape registration in the time of transformers.Advances in Neural Information Processing Systems, 34:5731–5744, 2021. 2

    2021

  44. [52]

    Nam: Neural adjoint maps for refining shape correspondences

    Giulio Vigan `o, Maks Ovsjanikov, and Simone Melzi. Nam: Neural adjoint maps for refining shape correspondences. ACM Trans. Graph., 44(4), 2025. 2, 5, 6

    2025

  45. [53]

    Bijective upsampling and learned embedding for point clouds correspondences.Com- puters and Graphics, 122:103985, 2024

    Giulio Vigan `o and Simone Melzi. Bijective upsampling and learned embedding for point clouds correspondences.Com- puters and Graphics, 122:103985, 2024. 2

    2024

  46. [54]

    Clip-nerf: Text-and-image driven manipula- tion of neural radiance fields.2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3825–3834, 2021

    Can Wang, Menglei Chai, Mingming He, Dongdong Chen, and Jing Liao. Clip-nerf: Text-and-image driven manipula- tion of neural radiance fields.2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3825–3834, 2021. 2 10

    2022

  47. [55]

    Romy Williamson and Niloy J. Mitra. Neural geometry pro- cessing via spherical neural surfaces.Computer Graphics Forum, 44(2):e70021, 2025. 2

    2025

  48. [56]

    Geometry processing with neural fields

    Guandao Yang, Serge Belongie, Bharath Hariharan, and Vladlen Koltun. Geometry processing with neural fields. In Thirty-Fifth Conference on Neural Information Processing Systems, 2021. 2

    2021

  49. [57]

    Geometry distribu- tions, 2024

    Biao Zhang, Jing Ren, and Peter Wonka. Geometry distribu- tions, 2024. 2, 5

    2024

  50. [58]

    Silvia Zuffi, Angjoo Kanazawa, David Jacobs, and Michael J. Black. 3D menagerie: Modeling the 3D shape and pose of animals. InProceedings IEEE Conference on Com- puter Vision and Pattern Recognition (CVPR) 2017, pages 5524–5532, Piscataway, NJ, USA, 2017. IEEE. 6 11

    2017

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.