pith. sign in

arxiv: 2605.24805 · v1 · pith:Z22XKYCQnew · submitted 2026-05-24 · 💻 cs.CV

Fishbone: From One 3D Asset to a Million Controllable Edits

Yumeng He , Xiaoying Wang , Peihao Li , Yanjia Huang , Joe Masterjohn , Jiajun Wu , Leonidas Guibas , Yin Yang
show 2 more authors
Ying Jiang Chenfanfu Jiang
This is my paper

Pith reviewed 2026-06-30 12:26 UTC · model grok-4.3

classification 💻 cs.CV
keywords 3D mesh deformationparametric editingrib-spine representationgeodesic scalar fieldcontrollable animationshape variationreduced-space dynamics
0
0 comments X

The pith

Fishbone turns any input 3D mesh into a rib-spine structure that lets ribs adjust local thickness and orientation while the spine governs global bends and twists for real-time deformation.

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

The paper presents Fishbone as a way to create controllable 3D assets from one mesh by automatically building a rib-spine control structure. It computes a geodesic scalar field on the mesh, pulls out iso-contours as ribs, threads a spine through their centers, and links surface points to these elements with Gaussian skinning weights. This setup lets users edit local cross-sections via rib parameters and global pose via spine parameters without hand-crafted rigs or per-category tuning. The same structure also drives reduced-space dynamics and keyframe animation, and the authors release a 136K-asset dataset built on top of it for downstream tasks in generation and robotics.

Core claim

Given an input mesh, Fishbone computes a geodesic scalar field with an adaptive heat method, extracts iso-contours as cross-sectional ribs, constructs a smooth geometry-aware spine through rib centers, and associates surface vertices with nearby rib and spine structures using Gaussian-weighted skinning. The resulting representation enables real-time and predictable deformation: ribs control local profiles such as thickness, orientation, and cross-sectional variation, while the spine controls global bending, twisting, and stretching.

What carries the argument

The rib-spine representation formed by iso-contour ribs from a geodesic scalar field, a spine through rib centers, and Gaussian-weighted skinning that ties mesh vertices to these controls.

Load-bearing premise

That the geodesic scalar field computed with the adaptive heat method, followed by iso-contour extraction and rib-center spine construction, will produce a meaningful and general-purpose control structure for arbitrary input meshes without category-specific tuning.

What would settle it

Running the extraction on a mesh with sharp mechanical features or thin handles and measuring whether scaling a single rib by 20 percent produces only the expected local thickness change without unintended global twisting or vertex collapse.

Figures

Figures reproduced from arXiv: 2605.24805 by Chenfanfu Jiang, Jiajun Wu, Joe Masterjohn, Leonidas Guibas, Peihao Li, Xiaoying Wang, Yanjia Huang, Ying Jiang, Yin Yang, Yumeng He.

Figure 1
Figure 1. Figure 1: Overview. We present Fishbone, a novel framework that constructs a coupled rib-spine representation from an input mesh and defines rib- and spine-driven deformation primitives. The proposed representation enables diverse and controllable shape variations, spanning local cross-sectional edits as well as global bending, twisting, and stretching, all driven by intuitive low-dimensional parameters. Moreover, t… view at source ↗
Figure 2
Figure 2. Figure 2: Rib-spine representation and shared-node spine tree. Left: Fish￾bone represents a mesh by surface ribs 𝑅𝑘 , shown as ordered iso-contour polylines, and spine key points p𝑘 , shown as rib centers connected into an orange spine. At branching levels, a single geodesic level may produce multi￾ple disconnected sub-ribs, such as 𝑅𝑘,1 and 𝑅𝑘,2. Right: the spine is stored as a shared-node tree: each spine S𝑏 visit… view at source ↗
Figure 3
Figure 3. Figure 3: Pipeline Overview. Starting from an input mesh M with one or multiple parts, we compute a geodesic field 𝜙 for each part using the adaptive heat method and extract ribs R as iso-contours of 𝜙. For each rib 𝑅𝑘 ∈ R, we select the locally flattest interior location as the representative spine point and connect them to form a branching spine S. We then compute per-vertex skinning weights W that bind the mesh t… view at source ↗
Figure 4
Figure 4. Figure 4: Deformation primitive gallery. Gallery of Fishbone parametric deformations applied to the bear mesh. Top two rows show six rib-driven variants produced by editing a single rib and propagating through projection￾based skinning (Eq. (7)): uniform scaling, anisotropic scaling, translation, rotation, local deformation, and cross-section reshape toward a square target. Bottom row shows the three spine-driven op… view at source ↗
Figure 5
Figure 5. Figure 5: Reduced simulation. We demonstrate visually plausible and inter￾active reduced-space dynamics with the proposed method: (i) Plant sway under wind: branch-aware stretch and length-normalized bending, com￾bined with root-pinned key points and a tip-biased wind ramp, produce coherent plant motion. The drag-form wind model adds turbulence and tangent-perpendicular projection to prevent stems from sliding along… view at source ↗
Figure 6
Figure 6. Figure 6: Keyframe animation. A rose blooming from bud to full bloom, authored as a few Fishbone edits and replayed by the per-vertex linear interpolation between keyframes (§4.5). where 𝜎imp controls the spatial support. Setting 𝜎imp → 0 applies the impulse only to the nearest key point. Moreover, general exter￾nal forces defined on the mesh surface, such as mesh-side contact, drag, ambient fields sampled at vertic… view at source ↗
Figure 7
Figure 7. Figure 7: Gallery. The proposed method generates stable and accurate rib-spine structures together with the corresponding skinning weights for diverse mesh inputs, without requiring per-asset tuning, category-specific templates, or manual rigging. Based on the generated representation, our framework further supports real-time, geometrically consistent deformation and efficient reduced-space dynamics simulation. part… view at source ↗
Figure 9
Figure 9. Figure 9: Dynamics latency scaling. Per-step latency scaling of the reduced￾dynamics pipeline. Left: latency vs. mesh vertex count 𝑁 , with linear fits reported in 𝜇s/vertex. The cylindrical lift’s per-vertex slope is roughly an order of magnitude steeper than the displacement lift’s because each vertex requires walking the parallel-transport frame and evaluating its rest cylin￾drical coordinates. Right: latency vs.… view at source ↗
Figure 10
Figure 10. Figure 10: Rib extraction design choices. Seven panels evaluate each component of our rib extraction on the same input mesh. Input mesh is the cleaned source. Ours runs the full pipeline: heat-method geodesic ribs with automatic root-axis selection and adaptive level count 𝐾 = clip(round(10 𝐿𝑝 /𝐿𝑜 ), 3, 10). The remaining five panels each remove one component: w/o heat method replaces the geodesic iso-contours with … view at source ↗
Figure 11
Figure 11. Figure 11: Score-maximization spine vs. midplane center. The midplane￾center baseline often drifts away from the natural interior of the shape and produces a kinked spine trajectory, whereas our method maintains a smooth and geometrically consistent spine within the object [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Branching tree handling. Effect of the BFS-based parent-child sub-rib tracking on multi-limb shapes. The proposed topology-aware branch￾ing tree connects sub-ribs across consecutive iso-levels through the face￾adjacency graph, enabling each branch spine to consistently follow its corresponding limb while sharing a single key point at Y-junctions. Without branching-tree handling, parent-child connectivity … view at source ↗
Figure 13
Figure 13. Figure 13: Mesh lifting. Effect of the two mesh-lift modes on a drag-induced bending deformation that pulls the bird’s head forward and downward. Input mesh: the rest pose with the user drag (red arrow) applied at the head. Cylindrical (Ours): per-vertex positions are reconstructed in the parallel￾transport frame of the deformed spine segment, so rib cross-sections rotate together with the bending spine and the head… view at source ↗
Figure 14
Figure 14. Figure 14: Image-conditioned mesh + Fishbone generation. The proposed method predicts the part-decomposed mesh and rib-spine structures from a single input image in a single forward pass. Compared with the baselines, the generated meshes exhibit cleaner part decomposition, more consistent geometric structures, and fewer floating artifacts. Metric PartCrafter [34] OmniPart [66] PartPacker [59] Ours CD ↓ 0.3085 0.2786… view at source ↗
Figure 15
Figure 15. Figure 15: Qualitative editing comparison. Side-by-side editing results of WIR3D [37], ARAP [53], and Fishbone. Fishbone provides explicit rib-spine handles that support localized and structured edits with lower manual setup. Zero-shot + Fishbone 𝚫 Category Single Multi Single Multi Single Multi Croissant 37.27 11.88 48.05 26.89 +10.77 +15.01 Tuna 21.72 6.800 26.89 24.17 +5.170 +17.38 Cup 87.42 61.17 89.49 72.76 +2.… view at source ↗
Figure 16
Figure 16. Figure 16: Dexterous grasping on in-the-wild scenes. Representative Dex￾GraspNet 2.0 [70] grasp executions on unseen in-the-wild meshes from three categories (rows, top to bottom: Croissant, Tuna, Cup). The left two columns show single-object grasp scenes before and after execution, while the right two columns show the corresponding multi-object grasp scenes with category-agnostic clutter surrounding the target. The… view at source ↗
Figure 17
Figure 17. Figure 17: Closed-loop Fishbone agent. The VLM observes a rendered screenshot, picks a tool from the tools in Tab. 5, applies it, and re-renders for the next round. An evaluator critic judges every deform call against the original request to gate retries. 6.3 Robot Learning via Data Diversification We use Fishbone as a geometry-aware data diversification frame￾work for dexterous grasping. Specifically, we adopt DexG… view at source ↗
Figure 18
Figure 18. Figure 18: Prompt Input Mesh "Make four legs thicker" Ours Agent Claude Code [PITH_FULL_IMAGE:figures/full_fig_p016_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Fishbone editor GUI. The interactive editor displays the input mesh together with its rib polylines and spine. The user picks a rib or spine-branch handle from the side panel and adjusts its parameters with sliders, invoking the deformation primitives of §3.5. Edits are applied at interactive rates via the pre-cached skinning matrix W (§3.4), with no per￾edit recomputation. F Agent Tools We list the ten-t… view at source ↗
read the original abstract

Large-scale controllable 3D assets are critical for computer graphics, embodied AI, robotics, and interactive content creation, yet creating diverse 3D assets remains challenging due to the high cost of manual modeling and rigging. Shape deformation offers a natural way to generate variations from existing meshes, but existing data-driven methods often rely on sparse user inputs, while parametric editing frameworks require manually designed control structures and category-specific configurations. Inspired by natural creatures, where a central spine governs global shape and cross-sectional ribs control local variation, we introduce Fishbone, a unified rib-spine representation for general shapes that supports controllable parametric mesh deformation, reduced-space dynamics, and animation. Given an input mesh, Fishbone computes a geodesic scalar field with an adaptive heat method, extracts iso-contours as cross-sectional ribs, constructs a smooth geometry-aware spine through rib centers, and associates surface vertices with nearby rib and spine structures using Gaussian-weighted skinning. The resulting representation enables real-time and predictable deformation: ribs control local profiles such as thickness, orientation, and cross-sectional variation, while the spine controls global bending, twisting, and stretching. The same structure also supports reduced-space simulation and keyframe animation. We further construct Fishbone-136K by augmenting Hunyuan3D with rib-spine structures, and demonstrate applications in controllable 3D generation, deformation-based data augmentation for robot learning, interactive mesh editing, and agentic generation. Experiments demonstrate the effectiveness, efficiency, and versatility of the proposed framework.

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

1 major / 0 minor

Summary. The paper introduces Fishbone, a rib-spine representation for general 3D meshes. Given an input mesh, it computes a geodesic scalar field via an adaptive heat method, extracts iso-contours as cross-sectional ribs, constructs a smooth spine through rib centers, and applies Gaussian-weighted skinning to associate vertices. Ribs control local thickness/orientation/variation while the spine controls global bending/twisting/stretching, enabling real-time predictable deformation, reduced-space simulation, and animation. The authors augment Hunyuan3D to create the Fishbone-136K dataset and demonstrate uses in controllable 3D generation, robot learning data augmentation, interactive editing, and agentic generation.

Significance. If the representation produces valid, non-intersecting ribs and a stable spine for arbitrary meshes, the approach would offer a notable advance over manual rigging or category-specific parametric models by providing an automatic, general-purpose control structure for deformation and animation. The large-scale dataset and breadth of demonstrated applications would further increase its utility in graphics, embodied AI, and robotics.

major comments (1)
  1. [Abstract] Abstract (pipeline description): the construction (adaptive heat method geodesic field → iso-contour ribs → rib-center spine → Gaussian skinning) is presented as applying to 'general shapes' and 'arbitrary input meshes,' yet the method implicitly assumes dominant tubular topology with a single source; no handling is described for branching, multiple medial axes, genus >0, or disconnected components. If iso-contours fail to close or the spine becomes ill-defined, the 'real-time and predictable deformation' guarantee does not hold without per-mesh tuning.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on the scope and assumptions of our method. We provide a point-by-point response below.

read point-by-point responses

  1. Referee: [Abstract] Abstract (pipeline description): the construction (adaptive heat method geodesic field → iso-contour ribs → rib-center spine → Gaussian skinning) is presented as applying to 'general shapes' and 'arbitrary input meshes,' yet the method implicitly assumes dominant tubular topology with a single source; no handling is described for branching, multiple medial axes, genus >0, or disconnected components. If iso-contours fail to close or the spine becomes ill-defined, the 'real-time and predictable deformation' guarantee does not hold without per-mesh tuning.

    Authors: The referee correctly identifies that our method relies on a dominant tubular topology with a single source for the geodesic field computation. The adaptive heat method and subsequent iso-contour extraction are designed under this assumption, as inspired by biological structures with a central spine. We do not claim or provide handling for branching topologies, multiple medial axes, genus greater than zero, or disconnected components in the current work. In cases where iso-contours fail to close or the spine is ill-defined, the deformation may require manual tuning. To address this, we will revise the abstract to remove the overgeneralization to 'arbitrary input meshes' and instead specify 'shapes with dominant tubular topology'. We will also add a limitations paragraph in the manuscript discussing these topological assumptions and potential failure modes. revision: yes

Circularity Check

0 steps flagged

No circularity: forward constructive pipeline from mesh to rib-spine structure

full rationale

The paper presents a procedural algorithm that takes an input mesh and computes a geodesic scalar field (adaptive heat method), extracts iso-contours as ribs, builds a spine from rib centers, and applies Gaussian skinning. No equations, fitted parameters, or predictions are defined in terms of the outputs; the central claim is a self-contained construction with no self-citation chains, ansatzes smuggled via prior work, or renaming of known results. The derivation chain does not reduce to its inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Only abstract available; no explicit free parameters, axioms, or invented entities beyond the named representation itself can be extracted.

invented entities (1)
  • Fishbone rib-spine representation no independent evidence
    purpose: Unified structure for controllable parametric deformation on general meshes
    New computational construct introduced by the paper; no independent evidence supplied in abstract.

pith-pipeline@v0.9.1-grok · 5831 in / 1068 out tokens · 26695 ms · 2026-06-30T12:26:17.515309+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

79 extracted references · 10 canonical work pages · 7 internal anchors

  1. [1]

    Hervé Abdi and Lynne J Williams. 2010. Principal component analysis.Wiley interdisciplinary reviews: computational statistics2, 4 (2010), 433–459

    2010

  2. [2]

    Ilge Akkaya, Marcin Andrychowicz, Maciek Chociej, Mateusz Litwin, Bob Mc- Grew, Arthur Petron, Alex Paino, Matthias Plappert, Glenn Powell, Raphael Ribas, et al. 2019. Solving rubik’s cube with a robot hand.arXiv preprint arXiv:1910.07113 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  3. [3]

    Nina Amenta, Sunghee Choi, and Ravi Krishna Kolluri. 2001. The power crust. InProceedings of the sixth ACM symposium on Solid modeling and applications. 249–266

    2001

  4. [4]

    Oscar Kin-Chung Au, Chiew-Lan Tai, Hung-Kuo Chu, Daniel Cohen-Or, and Tong-Yee Lee. 2008. Skeleton extraction by mesh contraction.ACM transactions on graphics (TOG)27, 3 (2008), 1–10

    2008

  5. [5]

    Ilya Baran and Jovan Popović. 2007. Automatic rigging and animation of 3d characters.ACM Transactions on graphics (TOG)26, 3 (2007), 72–es

    2007

  6. [6]

    Jernej Barbič and Doug L James. 2005. Real-time subspace integration for St. Venant-Kirchhoff deformable models.ACM transactions on graphics (TOG)24, 3 (2005), 982–990

    2005

  7. [7]

    Miklós Bergou, Max Wardetzky, Stephen Robinson, Basile Audoly, and Eitan Grinspun. 2008. Discrete elastic rods. InACM Siggraph 2008 Papers. 1–12

    2008

  8. [8]

    Volker Blanz and Thomas Vetter. 2023. A morphable model for the synthesis of 3D faces. InSeminal Graphics Papers: Pushing the Boundaries, Volume 2. 157–164

    2023

  9. [9]

    Sofien Bouaziz, Sebastian Martin, Tiantian Liu, Ladislav Kavan, and Mark Pauly

  10. [10]

    In Seminal Graphics Papers: Pushing the Boundaries, Volume 2

    Projective dynamics: Fusing constraint projections for fast simulation. In Seminal Graphics Papers: Pushing the Boundaries, Volume 2. 787–797

  11. [11]

    Hyeong In Choi, Sung Woo Choi, and Hwan Pyo Moon. 1997. Mathematical theory of medial axis transform.pacific journal of mathematics181, 1 (1997), 57–88

    1997

  12. [12]

    Min Gyu Choi and Hyeong-Seok Ko. 2005. Modal warping: Real-time simulation of large rotational deformation and manipulation.IEEE Transactions on Visualization & Computer Graphics11, 01 (2005), 91–101

    2005

  13. [13]

    Sabine Coquillart. 1990. Extended free-form deformation: A sculpturing tool for 3D geometric modeling. InProceedings of the 17th annual conference on Computer graphics and interactive techniques. 187–196

    1990

  14. [14]

    Keenan Crane, Clarisse Weischedel, and Max Wardetzky. 2013. Geodesics in heat: A new approach to computing distance based on heat flow.ACM Transactions on Graphics (ToG)32, 5 (2013), 1–11

    2013

  15. [15]

    Elisabetta Fedele, Francis Engelmann, Ian Huang, Or Litany, Marc Pollefeys, and Leonidas Guibas. 2025. SpaceControl: Introducing Test-Time Spatial Control to 3D Generative Modeling.arXiv preprint arXiv:2512.05343(2025)

    work page arXiv 2025

  16. [16]

    Mark Foskey, Maxim Garber, Ming C Lin, and Dinesh Manocha. 2001. A Voronoi- based hybrid motion planner. InProceedings 2001 IEEE/RSJ International Confer- ence on Intelligent Robots and Systems. Expanding the Societal Role of Robotics in the the Next Millennium (Cat. No. 01CH37180), Vol. 1. IEEE, 55–60

    2001

  17. [17]

    Lawson Fulton, Vismay Modi, David Duvenaud, David IW Levin, and Alec Jacob- son. 2019. Latent-space dynamics for reduced deformable simulation. InComputer graphics forum, Vol. 38. Wiley Online Library, 379–391

    2019

  18. [18]

    William Gao, Noam Aigerman, Thibault Groueix, Vova Kim, and Rana Hanocka

  19. [19]

    InACM SIG- GRAPH 2023 conference proceedings

    Textdeformer: Geometry manipulation using text guidance. InACM SIG- GRAPH 2023 conference proceedings. 1–11

    2023

  20. [20]

    Minghao Guo, Bohan Wang, and Wojciech Matusik. 2024. Medial skeletal diagram: A generalized medial axis approach for compact 3d shape representation.ACM Transactions on Graphics (TOG)43, 6 (2024), 1–23

    2024

  21. [21]

    Ernst Hairer, Marlis Hochbruck, Arieh Iserles, and Christian Lubich. 2006. Geo- metric numerical integration.Oberwolfach Reports3, 1 (2006), 805–882

    2006

  22. [22]

    Xiaoguang Han, Chang Gao, and Yizhou Yu. 2017. DeepSketch2Face: a deep learning based sketching system for 3D face and caricature modeling.ACM Transactions on graphics (TOG)36, 4 (2017), 1–12

    2017

  23. [23]

    Rana Hanocka, Noa Fish, Zhenhua Wang, Raja Giryes, Shachar Fleishman, and Daniel Cohen-Or. 2018. Alignet: Partial-shape agnostic alignment via unsuper- vised learning.ACM Transactions on Graphics (TOG)38, 1 (2018), 1–14

    2018

  24. [24]

    Hui Huang, Shihao Wu, Daniel Cohen-Or, Minglun Gong, Hao Zhang, Guiqing Li, and Baoquan Chen. 2013. L1-medial skeleton of point cloud.ACM Trans. Graph. 32, 4 (2013), 65–1

    2013

  25. [25]

    Takeo Igarashi, Satoshi Matsuoka, and Hidehiko Tanaka. 2006. Teddy: a sketching interface for 3D freeform design. InACM SIGGRAPH 2006 Courses. 11–es

    2006

  26. [26]

    Pushkar Joshi, Mark Meyer, Tony DeRose, Brian Green, and Tom Sanocki. 2007. Harmonic coordinates for character articulation.ACM transactions on graphics (TOG)26, 3 (2007), 71–es

    2007

  27. [27]

    Tao Ju, Scott Schaefer, and Joe Warren. 2023. Mean value coordinates for closed triangular meshes. InSeminal Graphics Papers: Pushing the Boundaries, Volume 2. 223–228

    2023

  28. [28]

    Heewoo Jun and Alex Nichol. 2023. Shap-e: Generating conditional 3d implicit functions.arXiv preprint arXiv:2305.02463(2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  29. [29]

    Ladislav Kavan, Steven Collins, Jiří Žára, and Carol O’Sullivan. 2007. Skinning with dual quaternions. InProceedings of the 2007 symposium on Interactive 3D graphics and games. 39–46

    2007

  30. [30]

    Theodore Kim and Doug L James. 2011. Physics-based character skinning us- ing multi-domain subspace deformations. InProceedings of the 2011 ACM SIG- GRAPH/eurographics symposium on computer animation. 63–72

    2011

  31. [31]

    Doris HU Kochanek and Richard H Bartels. 1984. Interpolating splines with local tension, continuity, and bias control. InProceedings of the 11th annual conference on Computer graphics and interactive techniques. 33–41

    1984

  32. [32]

    John Lasseter. 1998. Principles of traditional animation applied to 3D computer animation. InSeminal graphics: pioneering efforts that shaped the field. 263–272

    1998

  33. [33]

    Mingi Lee, Dongsu Zhang, Clément Jambon, and Young Min Kim. 2025. Brepdiff: Single-stage b-rep diffusion model. InProceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Papers. 1–11

    2025

  34. [34]

    Sergey Levine, Chelsea Finn, Trevor Darrell, and Pieter Abbeel. 2016. End-to-end training of deep visuomotor policies.Journal of Machine Learning Research17, 39 (2016), 1–40

    2016

  35. [35]

    John P Lewis, Matt Cordner, and Nickson Fong. 2023. Pose space deformation: a unified approach to shape interpolation and skeleton-driven deformation. In Seminal Graphics Papers: Pushing the Boundaries, Volume 2. 811–818

    2023

  36. [36]

    Yuchen Lin, Chenguo Lin, Panwang Pan, Honglei Yan, Yiqiang Feng, Yadong Mu, and Katerina Fragkiadaki. 2025. Partcrafter: Structured 3d mesh generation via compositional latent diffusion transformers.arXiv preprint arXiv:2506.05573 (2025)

    work page arXiv 2025

  37. [37]

    Yaron Lipman, David Levin, and Daniel Cohen-Or. 2008. Green coordinates.ACM transactions on graphics (TOG)27, 3 (2008), 1–10

    2008

  38. [38]

    Anran Liu, Cheng Lin, Yuan Liu, Xiaoxiao Long, Zhiyang Dou, Hao-Xiang Guo, Ping Luo, and Wenping Wang. 2024. Part123: Part-aware 3D Reconstruction from a Single-view Image. InACM SIGGRAPH 2024 Conference Papers. 1–12

    2024

  39. [39]

    Richard Liu, Daniel Fu, Noah Tan, Itai Lang, and Rana Hanocka. 2025. WIR3D: Visually-Informed and Geometry-Aware 3D Shape Abstraction. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    2025

  40. [40]

    Zhen Liu, Yao Feng, Michael J Black, Derek Nowrouzezahrai, Liam Paull, and Weiyang Liu. 2023. Meshdiffusion: Score-based generative 3d mesh modeling. arXiv preprint arXiv:2303.08133(2023)

    work page arXiv 2023

  41. [41]

    Matthew Loper, Naureen Mahmood, Javier Romero, Gerard Pons-Moll, and Michael J Black. 2023. SMPL: A skinned multi-person linear model. InSem- inal Graphics Papers: Pushing the Boundaries, Volume 2. 851–866

    2023

  42. [42]

    Shitong Luo and Wei Hu. 2021. Diffusion probabilistic models for 3d point cloud generation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2837–2845

    2021

  43. [43]

    Nadia Magnenat-Thalmann, Richard Laperrière, and Daniel Thalmann. 1989. Joint-dependent local deformations for hand animation and object grasping. In Proceedings on Graphics interface’88. 26–33

    1989

  44. [44]

    Mark Meyer, Mathieu Desbrun, Peter Schröder, and Alan H Barr. 2003. Discrete differential-geometry operators for triangulated 2-manifolds. InVisualization and mathematics III. Springer, 35–57

    2003

  45. [45]

    Niloy J Mitra, Michael Wand, Hao Zhang, Daniel Cohen-Or, Vladimir Kim, and Qi-Xing Huang. 2014. Structure-aware shape processing. InACM SIGGRAPH 2014 Courses. 1–21

    2014

  46. [46]

    Alex Nichol, Heewoo Jun, Prafulla Dhariwal, Pamela Mishkin, and Mark Chen

  47. [47]

    Point-E: A System for Generating 3D Point Clouds from Complex Prompts

    Point-e: A system for generating 3d point clouds from complex prompts. arXiv preprint arXiv:2212.08751(2022)

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  48. [48]

    Xue Bin Peng, Marcin Andrychowicz, Wojciech Zaremba, and Pieter Abbeel. 2018. Sim-to-real transfer of robotic control with dynamics randomization. In2018 IEEE international conference on robotics and automation (ICRA). IEEE, 3803–3810

    2018

  49. [49]

    Alex Pentland and John Williams. 1989. Good vibrations: Modal dynamics for graphics and animation. InProceedings of the 16th annual conference on Computer graphics and interactive techniques. 215–222

    1989

  50. [50]

    Ben Poole, Ajay Jain, Jonathan T Barron, and Ben Mildenhall. 2022. Dreamfusion: Text-to-3d using 2d diffusion.arXiv preprint arXiv:2209.14988(2022)

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  51. [51]

    1896.The theory of sound

    John William Strutt Baron Rayleigh. 1896.The theory of sound. Vol. 2. Macmillan

  52. [52]

    Fereshteh Sadeghi and Sergey Levine. 2016. Cad2rl: Real single-image flight without a single real image.arXiv preprint arXiv:1611.04201(2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  53. [53]

    Thomas W Sederberg and Scott R Parry. 1986. Free-form deformation of solid geometric models. InProceedings of the 13th annual conference on Computer graphics and interactive techniques. 151–160

    1986

  54. [54]

    James A Sethian. 1999. Fast marching methods.SIAM review41, 2 (1999), 199–235. 18•Yumeng He et al

    1999

  55. [55]

    Karan Singh and Eugene Fiume. 1998. Wires: a geometric deformation technique. InProceedings of the 25th annual conference on Computer graphics and interactive techniques. 405–414

    1998

  56. [56]

    Olga Sorkine, Marc Alexa, et al. 2007. As-rigid-as-possible surface modeling. In Symposium on Geometry processing, Vol. 4. 109–116

    2007

  57. [57]

    Vitaly Surazhsky, Tatiana Surazhsky, Danil Kirsanov, Steven J Gortler, and Hugues Hoppe. 2005. Fast exact and approximate geodesics on meshes.ACM transactions on graphics (TOG)24, 3 (2005), 553–560

    2005

  58. [58]

    Andrea Tagliasacchi, Ibraheem Alhashim, Matt Olson, and Hao Zhang. 2012. Mean curvature skeletons. InComputer Graphics Forum, Vol. 31. Wiley Online Library, 1735–1744

    2012

  59. [59]

    Andrea Tagliasacchi, Thomas Delame, Michela Spagnuolo, Nina Amenta, and Alexandru Telea. 2016. 3d skeletons: A state-of-the-art report. InComputer Graphics Forum, Vol. 35. Wiley Online Library, 573–597

    2016

  60. [60]

    Andrea Tagliasacchi, Hao Zhang, and Daniel Cohen-Or. 2009. Curve skeleton extraction from incomplete point cloud. InACM Siggraph 2009 Papers. 1–9

    2009

  61. [61]

    Qingyang Tan, Lin Gao, Yu-Kun Lai, Jie Yang, and Shihong Xia. 2018. Mesh-based autoencoders for localized deformation component analysis. InProceedings of the AAAI conference on artificial intelligence, Vol. 32

    2018

  62. [62]

    Jiaxiang Tang, Ruijie Lu, Max Li, Zekun Hao, Xuan Li, Fangyin Wei, Shuran Song, Gang Zeng, Ming-Yu Liu, and Tsung-Yi Lin. 2026. Efficient part-level 3d object generation via dual volume packing.Advances in Neural Information Processing Systems38 (2026), 27115–27137

    2026

  63. [63]

    Josh Tobin, Rachel Fong, Alex Ray, Jonas Schneider, Wojciech Zaremba, and Pieter Abbeel. 2017. Domain randomization for transferring deep neural networks from simulation to the real world. In2017 IEEE/RSJ international conference on intelligent robots and systems (IROS). IEEE, 23–30

    2017

  64. [64]

    Jakub Wejchert and David Haumann. 1991. Animation aerodynamics.ACM SIGGRAPH Computer Graphics25, 4 (1991), 19–22

    1991

  65. [65]

    Jiajun Wu, Chengkai Zhang, Tianfan Xue, Bill Freeman, and Josh Tenenbaum

  66. [66]

    Learning a probabilistic latent space of object shapes via 3d generative- adversarial modeling.Advances in neural information processing systems29 (2016)

    2016

  67. [67]

    Rundi Wu, Chang Xiao, and Changxi Zheng. 2021. Deepcad: A deep genera- tive network for computer-aided design models. InProceedings of the IEEE/CVF international conference on computer vision. 6772–6782

    2021

  68. [68]

    Jianfeng Xiang, Zelong Lv, Sicheng Xu, Yu Deng, Ruicheng Wang, Bowen Zhang, Dong Chen, Xin Tong, and Jiaolong Yang. 2024. Structured 3D Latents for Scalable and Versatile 3D Generation.arXiv preprint arXiv:2412.01506(2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  69. [69]

    Guandao Yang, Xun Huang, Zekun Hao, Ming-Yu Liu, Serge Belongie, and Bharath Hariharan. 2019. Pointflow: 3d point cloud generation with continuous normal- izing flows. InProceedings of the IEEE/CVF international conference on computer vision. 4541–4550

    2019

  70. [70]

    Yunhan Yang, Yufan Zhou, Yuan-Chen Guo, Zi-Xin Zou, Yukun Huang, Ying-Tian Liu, Hao Xu, Ding Liang, Yan-Pei Cao, and Xihui Liu. 2025. Omnipart: Part-aware 3d generation with semantic decoupling and structural cohesion. InProceedings of the SIGGRAPH Asia 2025 Conference Papers. 1–12

    2025

  71. [71]

    Wang Yifan, Noam Aigerman, Vladimir G Kim, Siddhartha Chaudhuri, and Olga Sorkine-Hornung. 2020. Neural cages for detail-preserving 3d deformations. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 75–83

    2020

  72. [72]

    Mehmet Ersin Yumer, Siddhartha Chaudhuri, Jessica K Hodgins, and Levent Burak Kara. 2015. Semantic shape editing using deformation handles.ACM Transactions on Graphics (TOG)34, 4 (2015), 1–12

    2015

  73. [73]

    M Ersin Yumer and Niloy J Mitra. 2016. Learning semantic deformation flows with 3d convolutional networks. InEuropean Conference on Computer Vision. Springer, 294–311

    2016

  74. [74]

    Jialiang Zhang, Haoran Liu, Danshi Li, XinQiang Yu, Haoran Geng, Yufei Ding, Jiayi Chen, and He Wang. 2024. DexGraspNet 2.0: Learning Generative Dexterous Grasping in Large-scale Synthetic Cluttered Scenes. In8th Annual Conference on Robot Learning

    2024

  75. [75]

    Jia-Peng Zhang, Cheng-Feng Pu, Meng-Hao Guo, Yan-Pei Cao, and Shi-Min Hu

  76. [76]

    One model to rig them all: Diverse skeleton rigging with unirig.ACM Transactions on Graphics (TOG)44, 4 (2025), 1–18

    2025

  77. [77]

    Zibo Zhao, Zeqiang Lai, Qingxiang Lin, Yunfei Zhao, Haolin Liu, Shuhui Yang, Yifei Feng, Mingxin Yang, Sheng Zhang, Xianghui Yang, et al. 2025. Hunyuan3d 2.0: Scaling diffusion models for high resolution textured 3d assets generation. arXiv preprint arXiv:2501.12202(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  78. [78]

    Xiangyu Zhu, Zhiqin Chen, Ruizhen Hu, and Xiaoguang Han. 2024. Control- lable shape modeling with neural generalized cylinder. InSIGGRAPH Asia 2024 Conference Papers. 1–11

    2024

  79. [79]

    Zeshun Zong, Xuan Li, Minchen Li, Maurizio M Chiaramonte, Wojciech Matusik, Eitan Grinspun, Kevin Carlberg, Chenfanfu Jiang, and Peter Yichen Chen. 2023. Neural stress fields for reduced-order elastoplasticity and fracture. InSIGGRAPH Asia 2023 Conference Papers. 1–11. Fishbone: From One 3D Asset to a Million Controllable Edits•19 A Mesh Preprocessing Thi...

    2023