Pith. sign in

REVIEW 3 major objections 6 minor 64 references

Generating Human-AI Collaborative Design Sequence for 3D Assets via Differentiable Operation Graph

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Optimizing a differentiable graph of 19 modeling operations against Chamfer distance recovers editable, human-aligned design sequences from arbitrary 3D assets without supervision.

desk verdict A promising but unverified approach to extracting editable 3D modeling histories; the headline fidelity claim is not backed because the exact extracted sequence is never evaluated, and the gated forward pass is undefined for topology-changing operations. read the letter →

arxiv 2508.17645 v2 pith:C2BTAACO submitted 2025-08-25 cs.GR

Xiaoyang Huang , Bingbing Ni , Wenjun Zhang This is my paper · ORCID
classification cs.GR MSC 68U05
keywords designoperationsequencesdifferentiableoperationsproceduralmodelingzero-shot3DreconstructionChamferdistanceoptimizationparametriceditinghuman-AIcollaborationmeshwiring
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 attempts to make AI-generated 3D content usable by professional designers by recovering, from any input 3D asset, a step-by-step construction history made of standard modeling operations. It reformulates 19 operations including Extrude, Bevel, Subdivision and Boolean as differentiable nodes, builds a hierarchical graph over them, and optimizes that graph per asset by minimizing Chamfer distance to the target shape, with penalties for invalid parameter combinations. The claimed outcome is zero-shot: no ground-truth operation sequences are needed, and the extracted sequences reconstruct the target accurately while remaining editable in standard digital content creation (DCC) software. If the claim holds, artists could edit AI outputs by adjusting parameters in the operation history rather than manipulating raw meshes, and the resulting models would have cleaner topology and edge flow for downstream animation and texturing. The paper reports Chamfer distance $0.823\times 10^{-2}$ on 503 assets and a normalized Longest Common Subsequence score of 0.942 against 52 human-built reference sequences.

What carries the argument

The central object is a hierarchical differentiable operation graph in which each node is one of 19 Blender-style modeling operations made differentiable. Continuous parameters flow through the chain rule; discrete parameters such as Boolean type, Bevel segment count, and face selection are handled by probabilistic branching, where parallel candidate outputs are averaged by Softmax weights; intrinsically discontinuous operations are given continuous surrogates, such as a fractional subdivision level $\beta_{su}$ that blends a smoothing subdivision with a topology-only subdivision, a UV-parameterized cylindrical sub-mesh for Extrude, an arc-based Bevel with continuously varying width, and a Boolean weight $\beta_{BL}$ that interpolates the loss between extruded and Boolean geometry. A gating coefficient per node interpolates between applying and skipping the operation, and the graph is optimized with Chamfer distance plus penalties that enforce compactness and domain rules. This machinery is what lets a purely geometric objective choose the operation sequence, its parameters, and even the sequence length without seeing a single ground-truth sequence.

What would settle it

Execute the extracted hard operation sequences exactly in a standard 3D application on the same 503 test assets and recompute Chamfer distance between the executed result and each target; if this number is materially worse than the reported $0.823\times 10^{-2}$, the measured accuracy belongs to the soft graph, not to the editable sequences.

Watch

Extended reading notes

Core claim

The central claim is that a design operation sequence for an arbitrary 3D asset can be extracted by gradient descent alone, without any dataset of human modeling histories. The paper constructs a graph whose nodes are differentiable versions of 19 modeling operations, each with continuous parameters, probabilistic-branch discrete choices, and a learnable gate that decides whether the operation is applied. Optimizing the graph end to end against Chamfer distance yields, at inference, a pruned chain of operations with concrete parameters, such as an Extrude followed by a Bevel followed by a Boolean. The paper reports that these chains achieve better geometric fidelity than neural mesh generation and CAD sequence reconstruction baselines (Chamfer distance $0.823\times 10^{-2}$ versus $2.857\times 10^{-2}$ for the best baseline) and that their order matches human design workflows on 52 manually constructed models. The point is not just reconstruction: because the sequence uses native software operations, every step remains editable and the mesh inherits the regular edge flow of procedural modeling.

Load-bearing premise

The paper's central numbers are measured on a soft, blended optimization; the method assumes the final exact sequence of operations achieves the same fidelity.

Editorial extensions

If this is right

  • Any existing 3D asset, including outputs from image-to-3D and text-to-3D generators, can be converted into a parametric construction history, so AI-generated content becomes editable in industry-standard software.
  • Artists can modify an extracted asset by changing one parameter, inserting an operation, or deleting a step, without re-modeling or global re-engineering.
  • The step-by-step decomposition doubles as an explanation of an asset's structure, usable for teaching, documentation, and design review.
  • Because the final meshes come from procedural operations, they inherit regular quad-dominant wiring, which supports animation deformation, simulation, and UV unwrapping better than typical neural mesh output.

Reading between the lines

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

  • A natural extension the paper does not explore is decomposing multi-part scenes by running the graph per connected component and then merging the resulting sub-sequences; the per-shape optimizer already localizes operations spatially, which may make this straightforward.
  • The close match to human sequences suggests the Chamfer-plus-penalty objective encodes a simplicity prior; deliberately varying the penalty strength should trade reconstruction error for even more human-like, shorter sequences, a testable knob the paper leaves open.
  • Since the optimizer takes about 20 minutes per asset on one GPU, warm-starting the graph from the solution of a nearby shape could turn it into an interactive reverse-engineering tool, but the paper does not report such transfer.
  • The fidelity of the extracted hard sequence versus the soft optimized mesh is measurable; if a gap appears, a fine-tuning pass that optimizes only the discrete sequence's parameters directly would be a natural fix.
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

3 major / 6 minor

Summary. The paper proposes a zero-shot method for extracting editable Blender-style design operation sequences from arbitrary 3D assets. It constructs a differentiable operation graph over 19 operations, with continuous parameters, discrete parameters handled by probabilistic branching, and a gating mechanism that interpolates between applying and skipping each operation. The graph is optimized per shape by minimizing Chamfer distance to the target geometry, and at inference the optimized graph is converted into a deterministic operation sequence by thresholding gate activations and taking argmax over discrete choices. The authors claim precise geometric reconstruction, high-quality mesh wiring, and alignment with human design workflows, and they support this with quantitative reconstruction metrics on 503 models, a comparison against 52 human-designed modeling sequences, mesh-wiring statistics, editing demonstrations, and a 407-participant user study.

Significance. If the technical gaps are closed, this is a potentially valuable step toward bridging neural 3D generation and parametric DCC workflows. The core idea is relevant and timely: representing AI-generated 3D content as editable procedural histories, rather than static meshes or neural fields, directly addresses a known production bottleneck. The paper also contains genuine strengths: the 52-model human-sequence benchmark is an external, non-synthetic evaluation, the differentiable reformulations of operations such as fractional subdivision and Boolean-via-Extrude are non-trivial, and the qualitative editing and integration demonstrations are appropriate for the claims made. The main risk is that the advertised geometric fidelity is not currently shown to describe the exact sequence a user would receive, which is the central promise of the method.

major comments (3)
  1. [Section III-B, Eq. (2)] The gated output x_{l+1} = {γ_l·N_l(x_l, θ_l, d_l), (1−γ_l)·x_l} is not defined as a mesh for topology-changing operations. For VertexDisplace and affine operations this is a convex combination of vertex positions, but for Extrude, Bevel, Boolean, and Subdivision the output of N_l generally has a different vertex set and connectivity from its input, and the paper supplies no vertex/face correspondence that would make the weighted combination a polygon mesh. Consequently the object whose Chamfer distance appears in Eqs. (3)–(4) is underspecified, and the geometric fidelity reported in Table III (CD 0.823e-2) is not shown to be the fidelity of an actual mesh. Please provide an explicit definition of the gated state for topology-changing nodes, for example a union of vertex sets with associated faces or an operation-specific interpolation, and verify that the reported loss is computed on that state.
  2. [Section IV-A, Table III] The headline reconstruction numbers are not shown to describe the extracted editable sequence. The final sequence is produced by retaining nodes with γ_l≥τ, taking argmax over discrete branches, and replacing βBL→1 Extrude surrogates with exact Boolean solvers (Sections III-B and III-Cd), but no experiment re-executes that exact sequence and recomputes CD, normal consistency, or F1. The text does not state whether the meshes used for Table III and rendered in Fig. 4 are soft-optimized graph outputs or re-executed exact sequences. Because the operation-label overlap in Table IV (LCS 14.3, Levenshtein Distance 1.4) compares symbols rather than resulting geometry, it does not establish that the exact sequence preserves the reported fidelity. Please add an explicit evaluation protocol in which the extracted sequence is executed exactly and all geometry metrics are recomputed on its output, and resolve the ambiguity in Fig. 4.
  3. [Appendix B / Table III] The learnable per-point visibility parameters z_i can mask geometry in the reported accuracy, and the paper does not state whether Table III uses masked or unmasked distances. In Appendix B, points whose visibility v_i=σ(z_i/τ_v) tends to zero are described as ignored in the loss, with only a λ_v∑v_i term preventing full collapse; if the Chamfer distance reported in Table III is computed through this visibility weighting, then CD 0.823e-2 may exclude hard-to-reconstruct regions. Please report the evaluation metric in unmasked form, or at minimum report both masked and unmasked CD, and state how the visibility weights are treated after optimization.
minor comments (6)
  1. [Section IV-A, Table III] The p-values and Cohen's h are not adequately described: with five runs per model it is unclear whether the statistical test is over runs, models, or both, and Cohen's h is designed for proportions rather than continuous effect sizes such as Chamfer distance.
  2. [Section III-Ca, Eq. (5)] The notation for the subdivision blend is confusing: Eq. (5) uses βSU while the fractional-level discussion uses βI_su, and the relationship between them is not stated. Please define a single interpolation parameter and use it consistently.
  3. [Section IV-Aa, Table IV] The sequence metrics compare operation labels only; please state explicitly whether parameter values such as Extrude heights or Bevel widths are also compared, and report the mean ground-truth sequence length so the values LCS 14.3 and Levenshtein Distance 1.4 can be interpreted.
  4. [Appendix B] The graph initialization uses C0=7 cycles while KnifeCut and LoopCut are said to be constrained to the initial 10 cycles; the mapping from cycles to operation positions in the extracted sequence should be clarified.
  5. [Figures 4, 6, and 8] The captions should state whether the displayed meshes come from re-executing the extracted sequence or from the soft-optimized graph, because this distinction is central to the geometric fidelity claim.
  6. [Section IV-A] The description of baseline inputs is unclear: DeepCAD and Point2CAD do not accept raw point clouds in their published forms, so the paper should specify how identical inputs were provided to each baseline.

Circularity Check

1 steps flagged · score 2.0 of 10

Geometric-fidelity evidence is the optimized Chamfer objective itself; the extracted exact sequence is never re-evaluated, but the 52-model LCS benchmark is external and keeps the central claim non-circular.

  1. fitted input called prediction [Sec. III-B (Eqs. 2-4) and Sec. IV-A (Table III)]
    "The graph is optimized end-to-end by minimizing the Chamfer Distance LCD between final outcome meshes and target shape: ... (3) ... Backpropagation simultaneously updates continuous parameters θ, discrete branch weights w, and gating weights ω. ... The final operation sequence is extracted by retaining nodes with γl≥τ and selecting discrete parameters through dl = arg max(pl), yielding a branch-free operation sequence with deterministic discrete parameters and optimal continuous parameters. ..."

    The mean CD reported in Table III is the value of the loss function minimized in Eqs. (3)-(4) on the same target geometry used for optimization, so the geometric-fidelity claim reduces to reporting the fit objective, not an independent prediction. The extraction step (thresholding γ, argmax discrete choices, replacing Boolean approximations) is not re-executed and re-measured, so the 0.823e-2 figure describes the soft blended graph rather than the delivered operation sequence. This makes the geometric validation partly tautological. The step-rationality claim, however, is checked against externally recorded human modeling sequences (52 models, Normalized LCS 0.942), which is not fitted, so the circularity is limited and does not infect the central sequence-alignment claim.

full rationale

The paper's core derivation—minimizing Chamfer Distance over a differentiable operation graph—is a per-shape inverse fitting procedure, and the reported CD numbers are the fitted objective values; that is a mild tautology inherent to reconstruction evaluation, not a hidden prediction. The independent 52-model human screen-recording benchmark provides external, non-fitted evidence for step rationality. The more serious concern is that the exact thresholded/argmax sequence is never executed and re-evaluated, but that is a validity gap, not a circular reduction. No load-bearing self-citation or imported uniqueness theorem appears; self-citations ([25]) are incidental related work. Overall score 2: one evaluation-metric tautology, central claim still has independent content.

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

The central claim depends on the expressive power of the operation vocabulary, the adequacy of Chamfer distance as a proxy for procedural rationality, and several underspecified mechanisms: the gated blend in Eq. 2, the soft-to-exact extraction step, the per-point visibility scheme, and the domain rule penalties. These are modeling choices or unstated assumptions rather than independently established facts, which is why the correctness risk is high despite the external step-rationality benchmark.

free parameters (6)
  • Per-point visibility parameters zi (with temperature τv=0.3) = per-shape, not reported
    Appendix B introduces a learnable visibility per output and target point; when vi→0 the point is ignored in the loss. If not tightly regularized, this can hide unmodeled geometry and lower the reported CD.
  • Activation pruning threshold τ = not reported
    Section III-B extracts the final sequence by retaining nodes with γl≥τ; the value determines the sequence length and geometry but is never given.
  • Initial graph cycles C0 and expansion schedule = C0=7
    Appendix B: the graph starts at 7 cycles and appends cycles upon stagnation; the thresholds controlling expansion and termination (ϵterm) are not specified.
  • Visibility regularizer weight λv = not reported
    Appendix B: λv Σvi prevents all vi→0; the value determines the strength of the geometric fidelity constraint and is not reported.
  • Alignment amplification ωalign = 5
    Appendix B: reweights well-aligned points by 5x during optimization; affects convergence and final geometry.
  • Bevel primary segment count K and later search range = K=5 primary; 1..9 search
    Section III-Cc fixes K=5 during primary optimization to reduce branching, then searches over 1..9; the final geometry depends on this schedule.
assumptions (6)
  • domain assumption The fixed set of 19 DCC operations, starting from a unit cube, can express all target shapes in the 503-model evaluation.
    Section III-A/B define the operation vocabulary and initial state. If a target requires an operation outside this set, the reported CD is unattainable; no expressiveness analysis is provided.
  • domain assumption Chamfer distance between sampled point clouds is a sufficient objective to recover semantically meaningful design sequences.
    Section III-B Eq. 3 uses CD as the sole geometric objective. Different sequences can yield the same mesh, and CD does not measure procedural rationality directly; step rationality is imposed via unspecified penalties.
  • ad hoc to paper The gated blend in Eq. 2 is well-defined for topology-changing operations.
    Eq. 2 writes x_{l+1} = {γ·N(x), (1−γ)·x}; operations like Extrude, Bevel, and Subdivision change the vertex set, so a convex combination of input and output meshes is not defined without a correspondence, which the paper does not provide.
  • ad hoc to paper The exact sequence extracted at inference reproduces the geometry of the soft-optimized blend.
    Section III-B extracts nodes with γl≥τ and resolves discrete parameters by argmax; Section III-Cd replaces approximated Booleans with exact ones. During training, γ and β weights are fractional, so re-executing exact operations can change geometry; no re-optimization or verification is described.
  • ad hoc to paper Learnable visibility parameters do not mask substantial geometry in the reported metrics.
    Appendix B introduces per-point visibility vi with weight λv; the paper does not state whether the reported CD uses full point clouds or visibility-weighted distances.
  • domain assumption The 52 models with human-recorded design sequences are representative of the 503-model test set.
    Section IV-Aa evaluates step rationality on 52 designer-built models; no selection criteria or distribution is reported, so the 90%+ alignment may not generalize.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generating Human-AI Collaborative Design Sequence for 3D Assets via Differentiable Operation Graph." pith.science (2026). https://pith.science/paper/C2BTAACO

@misc{pith2026250817645,
  author       = {Pith},
  title        = {Pith review of: Generating Human-AI Collaborative Design Sequence for 3D Assets via Differentiable Operation Graph},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C2BTAACO}},
  note         = {Machine review of arXiv:2508.17645}
}
read the original abstract

The emergence of 3D artificial intelligence-generated content (3D-AIGC) has enabled rapid synthesis of intricate geometries. However, a fundamental disconnect persists between AI-generated content and human-centric design paradigms, rooted in representational incompatibilities: conventional AI frameworks predominantly manipulate meshes or neural representations (\emph{e.g.}, NeRF, Gaussian Splatting), while designers operate within parametric modeling tools. This disconnection diminishes the practical value of AI for 3D industry, undermining the efficiency of human-AI collaboration. To resolve this disparity, we focus on generating design operation sequences, which are structured modeling histories that comprehensively capture the step-by-step construction process of 3D assets and align with designers' typical workflows in modern 3D software. We first reformulate fundamental modeling operations (\emph{e.g.}, \emph{Extrude}, \emph{Boolean}) into differentiable units, enabling joint optimization of continuous (\emph{e.g.}, \emph{Extrude} height) and discrete (\emph{e.g.}, \emph{Boolean} type) parameters via gradient-based learning. Based on these differentiable operations, a hierarchical graph with gating mechanism is constructed and optimized end-to-end by minimizing Chamfer Distance to target geometries. Multi-stage sequence length constraint and domain rule penalties enable unsupervised learning of compact design sequences without ground-truth sequence supervision. Extensive validation demonstrates that the generated operation sequences achieve high geometric fidelity, smooth mesh wiring, rational step composition and flexible editing capacity, with full compatibility within design industry.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 47 canonical work pages

  1. [1]

    Dreamflow: High-quality text-to-3d generation by approximating probability flow,

    K. Lee, K. Sohn, and J. Shin, “Dreamflow: High-quality text-to-3d generation by approximating probability flow,” in 12th International Conference on Learning Representations, ICLR 2024 , 2024

    work page 2024

  2. [2]

    Hyperfields: Towards zero-shot generation of nerfs from text,

    S. Babu, R. Liu, A. Zhou, M. Maire, G. Shakhnarovich, and R. Hanocka, “Hyperfields: Towards zero-shot generation of nerfs from text,” in International Conference on Machine Learning . PMLR, 2024, pp. 2230–2247

    work page 2024

  3. [3]

    Shapecoder: Dis- covering abstractions for visual programs from unstructured primitives,

    R. K. Jones, P. Guerrero, N. J. Mitra, and D. Ritchie, “Shapecoder: Dis- covering abstractions for visual programs from unstructured primitives,” ACM Transactions on Graphics (TOG) , 2023

    work page 2023

  4. [4]

    Shapemod: macro operation discovery for 3d shape programs,

    R. K. Jones, D. Charatan, P. Guerrero, N. J. Mitra, and D. Ritchie, “Shapemod: macro operation discovery for 3d shape programs,” ACM Transactions on Graphics (TOG) , vol. 40, no. 4, pp. 1–16, 2021

    work page 2021

  5. [5]

    Shapeassembly: Learning to generate programs for 3d shape structure synthesis,

    R. K. Jones, T. Barton, X. Xu, K. Wang, E. Jiang, P. Guerrero, N. J. Mitra, and D. Ritchie, “Shapeassembly: Learning to generate programs for 3d shape structure synthesis,”ACM Transactions on Graphics (TOG), vol. 39, no. 6, pp. 1–20, 2020

    work page 2020

  6. [6]

    Infinite photorealistic worlds using procedural generation,

    A. Raistrick, L. Lipson, Z. Ma, L. Mei, M. Wang, Y . Zuo, K. Kayan, H. Wen, B. Han, Y . Wang et al. , “Infinite photorealistic worlds using procedural generation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2023, pp. 12 630–12 641

    work page 2023

  7. [7]

    Infinigen indoors: Photorealistic indoor scenes using procedural generation,

    A. Raistrick, L. Mei, K. Kayan, D. Yan, Y . Zuo, B. Han, H. Wen, M. Parakh, S. Alexandropoulos, L. Lipson et al. , “Infinigen indoors: Photorealistic indoor scenes using procedural generation,” in Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 21 783–21 794

    work page 2024

  8. [8]

    Blenderalchemy: Editing 3d graphics with vision-language models,

    I. Huang, G. Yang, and L. Guibas, “Blenderalchemy: Editing 3d graphics with vision-language models,” in European Conference on Computer Vision. Springer, 2024, pp. 297–314

    2024

Show all 64 references
  1. [9]

    Blendergym: Benchmarking foundational model systems for graphics editing,

    Y . Gu, I. Huang, J. Je, G. Yang, and L. Guibas, “Blendergym: Benchmarking foundational model systems for graphics editing,” arXiv preprint arXiv:2504.01786, 2025

    2025 arXiv

  2. [10]

    Meshanything: Artist-created mesh generation with autoregressive transformers,

    Y . Chen, T. He, D. Huang, W. Ye, S. Chen, J. Tang, X. Chen, Z. Cai, L. Yang, G. Yu et al. , “Meshanything: Artist-created mesh generation with autoregressive transformers,” arXiv preprint arXiv:2406.10163 , 2024

    2024 arXiv

  3. [11]

    Meshanything v2: Artist-created mesh generation with adjacent mesh tokenization,

    Y . Chen, Y . Wang, Y . Luo, Z. Wang, Z. Chen, J. Zhu, C. Zhang, and G. Lin, “Meshanything v2: Artist-created mesh generation with adjacent mesh tokenization,” arXiv preprint arXiv:2408.02555 , 2024

    2024 arXiv

  4. [12]

    Meshgpt: Generating triangle meshes with decoder-only transformers,

    Y . Siddiqui, A. Alliegro, A. Artemov, T. Tommasi, D. Sirigatti, V . Rosov, A. Dai, and M. Nießner, “Meshgpt: Generating triangle meshes with decoder-only transformers,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2024, pp. 19 615–19 625

    2024

  5. [13]

    Nerf: Representing scenes as neural radiance fields for view synthesis,

    B. Mildenhall, P. P. Srinivasan, M. Tancik, J. T. Barron, R. Ramamoorthi, and R. Ng, “Nerf: Representing scenes as neural radiance fields for view synthesis,” Communications of the ACM , vol. 65, no. 1, pp. 99–106, 2021

    2021

  6. [14]

    Dreamfusion: Text- to-3d using 2d diffusion,

    B. Poole, A. Jain, J. T. Barron, and B. Mildenhall, “Dreamfusion: Text- to-3d using 2d diffusion,” in The Eleventh International Conference on Learning Representations

  7. [15]

    3d gaussian splatting for real-time radiance field rendering

    B. Kerbl, G. Kopanas, T. Leimk ¨uhler, and G. Drettakis, “3d gaussian splatting for real-time radiance field rendering.” ACM Trans. Graph. , vol. 42, no. 4, pp. 139–1, 2023

    2023

  8. [16]

    Gaussiandreamer: Fast generation from text to 3d gaussians by bridging 2d and 3d diffusion models,

    T. Yi, J. Fang, J. Wang, G. Wu, L. Xie, X. Zhang, W. Liu, Q. Tian, and X. Wang, “Gaussiandreamer: Fast generation from text to 3d gaussians by bridging 2d and 3d diffusion models,” in CVPR, 2024

    2024

  9. [17]

    Cad-signet: Cad language inference from point clouds using layer-wise sketch instance guided attention,

    M. S. Khan, E. Dupont, S. A. Ali, K. Cherenkova, A. Kacem, and D. Aouada, “Cad-signet: Cad language inference from point clouds using layer-wise sketch instance guided attention,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 4713–4722

    2024

  10. [18]

    Extrudenet: Unsupervised inverse sketch-and-extrude for shape parsing,

    D. Ren, J. Zheng, J. Cai, J. Li, and J. Zhang, “Extrudenet: Unsupervised inverse sketch-and-extrude for shape parsing,” in European Conference on Computer Vision . Springer, 2022, pp. 482–498

    2022

  11. [19]

    Skexgen: Autoregressive generation of cad construction sequences with disentangled codebooks,

    X. Xu, K. D. Willis, J. G. Lambourne, C.-Y . Cheng, P. K. Jayaraman, and Y . Furukawa, “Skexgen: Autoregressive generation of cad construction sequences with disentangled codebooks,” in International Conference on Machine Learning . PMLR, 2022, pp. 24 698–24 724

    2022

  12. [20]

    Split- and-fit: Learning b-reps via structure-aware voronoi partitioning,

    Y . Liu, J. Chen, S. Pan, D. Cohen-Or, H. Zhang, and H. Huang, “Split- and-fit: Learning b-reps via structure-aware voronoi partitioning,” ACM Transactions on Graphics (TOG) , vol. 43, no. 4, pp. 1–13, 2024

    2024

  13. [21]

    Free2cad: Parsing freehand drawings into cad commands,

    C. Li, H. Pan, A. Bousseau, and N. J. Mitra, “Free2cad: Parsing freehand drawings into cad commands,” ACM Transactions on Graphics (TOG) , vol. 41, no. 4, pp. 1–16, 2022

    2022

  14. [22]

    Vitruvion: A gen- erative model of parametric cad sketches,

    A. Seff, W. Zhou, N. Richardson, and R. P. Adams, “Vitruvion: A gen- erative model of parametric cad sketches,” in International Conference on Learning Representations , 2021

    2021

  15. [23]

    Capri-net: Learning compact cad shapes with adaptive primitive assembly,

    F. Yu, Z. Chen, M. Li, A. Sanghi, H. Shayani, A. Mahdavi-Amiri, and H. Zhang, “Capri-net: Learning compact cad shapes with adaptive primitive assembly,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 11 768–11 778

    2022

  16. [24]

    3dias: 3d shape reconstruction with implicit algebraic surfaces,

    M. Yavartanoo, J. Chung, R. Neshatavar, and K. M. Lee, “3dias: 3d shape reconstruction with implicit algebraic surfaces,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 12 446–12 455

    2021

  17. [25]

    Learning shape primitives via implicit convexity regularization,

    X. Huang, Y . Zhang, K. Chen, T. Li, W. Zhang, and B. Ni, “Learning shape primitives via implicit convexity regularization,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 3642–3651

    2023

  18. [26]

    Bsp-net: Generating compact meshes via binary space partitioning,

    Z. Chen, A. Tagliasacchi, and H. Zhang, “Bsp-net: Generating compact meshes via binary space partitioning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 45– 54

    2020

  19. [27]

    Bae-net: Branched autoencoder for shape co-segmentation,

    Z. Chen, K. Yin, M. Fisher, S. Chaudhuri, and H. Zhang, “Bae-net: Branched autoencoder for shape co-segmentation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2019, pp. 8490–8499. 17

    2019

  20. [28]

    Robust and accurate superquadric recovery: a probabilistic approach,

    W. Liu, Y . Wu, S. Ruan, and G. S. Chirikjian, “Robust and accurate superquadric recovery: a probabilistic approach,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 2676–2685

    2022

  21. [29]

    Cvxnet: Learnable convex decomposition,

    B. Deng, K. Genova, S. Yazdani, S. Bouaziz, G. Hinton, and A. Tagliasacchi, “Cvxnet: Learnable convex decomposition,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 31–44

    2020

  22. [30]

    Learning unsupervised hierarchical part decomposition of 3d objects from a single rgb image,

    D. Paschalidou, L. V . Gool, and A. Geiger, “Learning unsupervised hierarchical part decomposition of 3d objects from a single rgb image,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 1060–1070

    2020

  23. [31]

    Superquadrics revisited: Learning 3d shape parsing beyond cuboids,

    D. Paschalidou, A. O. Ulusoy, and A. Geiger, “Superquadrics revisited: Learning 3d shape parsing beyond cuboids,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 10 344–10 353

    2019

  24. [32]

    Neural star domain as primitive representation,

    Y . Kawana, Y . Mukuta, and T. Harada, “Neural star domain as primitive representation,” Advances in Neural Information Processing Systems , vol. 33, pp. 7875–7886, 2020

    2020

  25. [33]

    Neural parts: Learning expressive 3d shape abstractions with invertible neural networks,

    D. Paschalidou, A. Katharopoulos, A. Geiger, and S. Fidler, “Neural parts: Learning expressive 3d shape abstractions with invertible neural networks,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 3204–3215

    2021

  26. [34]

    Topology guaranteed b-spline surface/surface intersection,

    J. Yang, X. Jia, and D.-M. Yan, “Topology guaranteed b-spline surface/surface intersection,” ACM Transactions on Graphics (TOG) , vol. 42, no. 6, pp. 1–16, 2023

    2023

  27. [35]

    Sketch-based generation and editing of quad meshes

    K. Takayama, D. Panozzo, A. Sorkine-Hornung, and O. Sorkine- Hornung, “Sketch-based generation and editing of quad meshes.” ACM Trans. Graph., vol. 32, no. 4, pp. 97–1, 2013

    2013

  28. [36]

    Data-driven interactive quadrangulation,

    G. Marcias, K. Takayama, N. Pietroni, D. Panozzo, O. Sorkine-Hornung, E. Puppo, and P. Cignoni, “Data-driven interactive quadrangulation,” ACM Transactions on Graphics (TOG) , vol. 34, no. 4, pp. 1–10, 2015

    2015

  29. [37]

    Super-resolution cloth animation with spatial and temporal coherence,

    J. Yu and Z. Wang, “Super-resolution cloth animation with spatial and temporal coherence,” ACM Transactions on Graphics (TOG) , vol. 43, no. 4, pp. 1–14, 2024

    2024

  30. [38]

    Skinmixer: Blending 3d animated models,

    S. Nuvoli, N. Pietroni, P. Cignoni, R. Scateni, and M. Tarini, “Skinmixer: Blending 3d animated models,” ACM Transactions on Graphics (TOG) , vol. 41, no. 6, pp. 1–15, 2022

    2022

  31. [39]

    Regularized kelvinlets: sculpting brushes based on fundamental solutions of elasticity,

    F. De Goes and D. L. James, “Regularized kelvinlets: sculpting brushes based on fundamental solutions of elasticity,” ACM Transactions on Graphics (TOG), vol. 36, no. 4, pp. 1–11, 2017

    2017

  32. [40]

    Joint uv optimization and texture baking,

    J. Knodt, Z. Pan, K. Wu, and X. Gao, “Joint uv optimization and texture baking,” ACM Transactions on Graphics, vol. 43, no. 1, pp. 1–20, 2023

    2023

  33. [41]

    Abc: A big cad model dataset for geometric deep learning,

    S. Koch, A. Matveev, Z. Jiang, F. Williams, A. Artemov, E. Burnaev, M. Alexa, D. Zorin, and D. Panozzo, “Abc: A big cad model dataset for geometric deep learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 9601–9611

    2019

  34. [42]

    Deepcad: A deep generative network for computer-aided design models,

    R. Wu, C. Xiao, and C. Zheng, “Deepcad: A deep generative network for computer-aided design models,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 6772–6782

    2021

  35. [43]

    Cad-mllm: Unifying multimodality-conditioned cad generation with mllm,

    J. Xu, C. Wang, Z. Zhao, W. Liu, Y . Ma, and S. Gao, “Cad-mllm: Unifying multimodality-conditioned cad generation with mllm,” arXiv preprint arXiv:2411.04954, 2024

    2024 arXiv

  36. [44]

    Get3d: A generative model of high quality 3d textured shapes learned from images,

    J. Gao, T. Shen, Z. Wang, W. Chen, K. Yin, D. Li, O. Litany, Z. Gojcic, and S. Fidler, “Get3d: A generative model of high quality 3d textured shapes learned from images,” Advances In Neural Information Processing Systems, vol. 35, pp. 31 841–31 854, 2022

    2022

  37. [45]

    Efficient geometry-aware 3d generative adversarial networks,

    E. R. Chan, C. Z. Lin, M. A. Chan, K. Nagano, B. Pan, S. De Mello, O. Gallo, L. J. Guibas, J. Tremblay, S. Khamis et al. , “Efficient geometry-aware 3d generative adversarial networks,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022,...

    2022

  38. [46]

    Graf: Generative radiance fields for 3d-aware image synthesis,

    K. Schwarz, Y . Liao, M. Niemeyer, and A. Geiger, “Graf: Generative radiance fields for 3d-aware image synthesis,” Advances in Neural Information Processing Systems , vol. 33, pp. 20 154–20 166, 2020

    2020

  39. [47]

    Representing scenes as compositional generative neural feature fields based on giraffe for 3d reconstruction of classroom scenes,

    T. Fu, R. Deng, Y . Gao, and F. Zhang, “Representing scenes as compositional generative neural feature fields based on giraffe for 3d reconstruction of classroom scenes,” in International Conference on Intelligent Information Hiding and Multimedia Signal Processing . Springer,...

    2022

  40. [48]

    Dreamfusion: Text- to-3d using 2d diffusion,

    B. Poole, A. Jain, J. T. Barron, and B. Mildenhall, “Dreamfusion: Text- to-3d using 2d diffusion,” arXiv preprint arXiv:2209.14988 , 2022

    2022 arXiv

  41. [49]

    Point-e: A system for generating 3d point clouds from complex prompts,

    A. Nichol, H. Jun, P. Dhariwal, P. Mishkin, and M. Chen, “Point-e: A system for generating 3d point clouds from complex prompts,” arXiv preprint arXiv:2212.08751, 2022

    2022 arXiv

  42. [50]

    3d neural field generation using triplane diffusion,

    J. R. Shue, E. R. Chan, R. Po, Z. Ankner, J. Wu, and G. Wetzstein, “3d neural field generation using triplane diffusion,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 20 875–20 886

    2023

  43. [51]

    Rodin: A generative model for sculpting 3d digital avatars using diffusion,

    T. Wang, B. Zhang, T. Zhang, S. Gu, J. Bao, T. Baltrusaitis, J. Shen, D. Chen, F. Wen, Q. Chen et al. , “Rodin: A generative model for sculpting 3d digital avatars using diffusion,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp...

    2023

  44. [52]

    Hunyuan3d 2.0: Scaling diffusion mod- els for high resolution textured 3d assets generation,

    Z. Zhao, Z. Lai, Q. Lin, Y . Zhao, H. Liu, S. Yang, Y . Feng, M. Yang, S. Zhang, X. Yang et al. , “Hunyuan3d 2.0: Scaling diffusion mod- els for high resolution textured 3d assets generation,” arXiv preprint arXiv:2501.12202, 2025

    2025 arXiv

  45. [53]

    Structured 3d latents for scalable and versatile 3d generation,

    J. Xiang, Z. Lv, S. Xu, Y . Deng, R. Wang, B. Zhang, D. Chen, X. Tong, and J. Yang, “Structured 3d latents for scalable and versatile 3d generation,” arXiv preprint arXiv:2412.01506 , 2024

    2024 arXiv

  46. [54]

    Hyper3d: Ai-powered 3d content generation platform,

    Hyper3D Team, “Hyper3d: Ai-powered 3d content generation platform,” https://hyper3d.ai, 2025, accessed: 2025-04-06

    2025

  47. [55]

    Ucsg-net-unsupervised discovering of constructive solid geometry tree,

    K. Kania, M. Zieba, and T. Kajdanowicz, “Ucsg-net-unsupervised discovering of constructive solid geometry tree,” Advances in Neural Information Processing Systems , vol. 33, pp. 8776–8786, 2020

    2020

  48. [56]

    Dualcsg: Learning dual csg trees for general and compact cad modeling,

    F. Yu, Q. Chen, M. Tanveer, A. M. Amiri, and H. Zhang, “Dualcsg: Learning dual csg trees for general and compact cad modeling,” arXiv preprint arXiv:2301.11497, 2023

    2023 arXiv

  49. [57]

    Mesh arrangements for solid geometry,

    Q. Zhou, E. Grinspun, D. Zorin, and A. Jacobson, “Mesh arrangements for solid geometry,” ACM Transactions on Graphics (TOG) , vol. 35, no. 4, pp. 1–15, 2016

    2016

  50. [58]

    Point2cad: Reverse engineering cad models from 3d point clouds,

    Y . Liu, A. Obukhov, J. D. Wegner, and K. Schindler, “Point2cad: Reverse engineering cad models from 3d point clouds,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 3763–3772

    2024

  51. [59]

    Blenderkit – 3d asset repository,

    BlenderKit, “Blenderkit – 3d asset repository,” 2025, accessed: 2025-01-20. [Online]. Available: https://www.blenderkit.com/

    2025

  52. [60]

    Free3d – download free 3d models,

    Free3D, “Free3d – download free 3d models,” 2025, accessed: 2025-01-20. [Online]. Available: https://free3d.com/

    2025

  53. [61]

    Sketchfab – 3d & ar visualization platform,

    Sketchfab, “Sketchfab – 3d & ar visualization platform,” 2025, accessed: 2025-01-20. [Online]. Available: https://sketchfab.com/

    2025

  54. [62]

    Parsenet: A parametric surface fitting network for 3d point clouds,

    G. Sharma, D. Liu, S. Maji, E. Kalogerakis, S. Chaudhuri, and R. M ˇech, “Parsenet: A parametric surface fitting network for 3d point clouds,” in European Conference on Computer Vision . Springer, 2020, pp. 261– 276

    2020

  55. [63]

    Gaussianeditor: Swift and controllable 3d editing with gaussian splatting,

    Y . Chen, Z. Chen, C. Zhang, F. Wang, X. Yang, Y . Wang, Z. Cai, L. Yang, H. Liu, and G. Lin, “Gaussianeditor: Swift and controllable 3d editing with gaussian splatting,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2024, pp. 21 476–21 485

    2024

  56. [64]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in ICLR, 2014. 18 APPENDIX A ELABORATION ON OPERATION PARAMETERS Subdivision. Subdivision level is a non-negative integer (Z≥0) indicating the number of surface subdivision iterations. A level of 0 means no ...

    2014

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.