Pith. sign in

REVIEW 2 major objections 1 minor 43 references

Self-supervised Garment Dynamics with Persistent Wrinkles

T0 review · 2 major / 1 minor · reviewed 2026-06-30 · grok-4.3

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

Pith's one-line read A self-supervised neural simulator produces persistent garment wrinkles by turning energy minimization into a moving problem that mimics plasticity.

desk verdict The paper's new piece is a curriculum that ramps the loss target from elastic to elasto-plastic so a self-supervised garment network can produce persistent wrinkles, but the moving-objective convergence has no analysis or ablations to back it. read the letter →

arxiv 2606.25065 v3 pith:RFZOQDKY submitted 2026-06-23 cs.GR

Xiaoyuan Yang , Deshan Gong , Taku Komura , He Wang This is my paper · ORCID
classification cs.GR
keywords self-supervisedlearninggarmentsimulationpersistentwrinkleselasto-plasticitycurriculumneuralclothphysics-inspiredlossdynamics
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 aims to show that self-supervised neural networks can simulate garment dynamics with realistic, lasting wrinkles without any training data. Existing approaches simplify cloth to pure elasticity and therefore cannot produce believable folds that remain after deformation. The authors address this by defining a physics-inspired loss that makes the optimization target change over time to capture plastic behavior. Training stability is achieved through a curriculum that starts the network on elastic materials and gradually introduces plasticity so that the loss and the network parameters converge together. If correct, this means self-supervised cloth models can reach higher visual quality while retaining their data-free and fast-inference advantages.

What carries the argument

A physics-inspired loss function that converts learning into a moving energy minimization problem to mimic plasticity, stabilized by a curriculum that gradually shifts the target material from pure elasticity to elasto-plasticity.

What would settle it

Train the network without the curriculum and check whether it either diverges or produces only transient wrinkles that vanish once the body stops moving.

Watch

Extended reading notes

Core claim

The central claim is that the first self-supervised neural garment simulator to explicitly model persistent wrinkles can be built by using a novel physics-inspired loss function that turns learning into a moving energy minimization problem to mimic plasticity, together with a curriculum learning scheme in which the target material gradually changes from pure elasticity to elasto-plasticity so that the loss function and the learnable parameters jointly converge.

Load-bearing premise

A curriculum learning scheme in which the target material gradually changes from pure elasticity to elasto-plasticity allows the loss function and the learnable parameters to jointly converge.

Editorial extensions

If this is right

  • Self-supervised models can now generate natural persistent wrinkles across a range of garments, body shapes, and motions.
  • Visual realism improves according to multiple quantitative metrics while preserving the efficiency and data-free nature of the approach.
  • The same curriculum strategy enables joint convergence when the loss itself changes during optimization.
  • The resulting simulator works on both static and dynamic sequences without requiring supervised training pairs.

Reading between the lines

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

  • The curriculum idea could be tested on other self-supervised physics tasks where the objective must change from simple to complex behavior.
  • One could measure whether the learned wrinkles match real fabric by comparing simulated sequences to video of actual cloth that has been creased and then released.
  • The method suggests a route to add plastic effects to other neural simulators such as hair or soft-body animation without collecting new data.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 1 minor

Summary. The paper claims to introduce the first self-supervised neural garment simulator that explicitly models persistent wrinkles caused by plasticity. This is achieved through a novel physics-inspired loss function that turns learning into a moving energy minimization problem to mimic plasticity. To handle the resulting training difficulties with a changing loss, the authors propose a physics-inspired curriculum learning scheme in which the target material gradually changes from pure elasticity to elasto-plasticity, allowing the loss function and learnable parameters to jointly converge. Comprehensive evaluations are said to demonstrate that the method generates natural persistent wrinkles and outperforms existing methods across garments, body shapes, and motions according to a range of metrics.

Significance. If the curriculum successfully stabilizes convergence under the moving objective, the result would be significant for neural garment simulation by extending self-supervised methods (which require no training data) to elasto-plastic behaviors that produce believable persistent wrinkles. The physics-inspired formulation and data-free training are explicit strengths that could improve visual realism in animation pipelines.

major comments (2)
  1. [Curriculum Learning Scheme] The curriculum learning scheme (described in the abstract and presumably detailed in the methods section) is load-bearing for the central claim that persistent wrinkles can be obtained self-supervised. However, the manuscript supplies no schedule details (rate of plasticity-parameter ramp, number of stages, or functional form), no convergence analysis, and no ablation showing that abrupt or absent curriculum produces divergence or collapsed wrinkles. This directly tests whether the optimizer can track the moving minimum as the loss changes during optimization.
  2. [Evaluation] The evaluation section claims the method 'outperforms existing methods on a variety of garments, body shapes, and body motions, according to a range of metrics,' yet the abstract provides no quantitative results, specific metrics for wrinkle persistence, or baseline comparisons. Without these, it is impossible to verify whether the persistent-wrinkles claim holds or reduces to visual inspection.
minor comments (1)
  1. [Abstract] The abstract asserts the approach is 'the first' self-supervised simulator with persistent wrinkles; this novelty claim should be supported by a clear comparison table or related-work discussion in the main text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The two major comments highlight important aspects of the curriculum learning scheme and the presentation of evaluation results. We address each point below and indicate where revisions will be made to strengthen the paper.

read point-by-point responses

  1. Referee: [Curriculum Learning Scheme] The curriculum learning scheme (described in the abstract and presumably detailed in the methods section) is load-bearing for the central claim that persistent wrinkles can be obtained self-supervised. However, the manuscript supplies no schedule details (rate of plasticity-parameter ramp, number of stages, or functional form), no convergence analysis, and no ablation showing that abrupt or absent curriculum produces divergence or collapsed wrinkles. This directly tests whether the optimizer can track the moving minimum as the loss changes during optimization.

    Authors: We agree that explicit details on the curriculum are necessary to substantiate the central claim. The current manuscript describes the gradual transition from elastic to elasto-plastic targets at a high level but does not provide the precise ramp schedule, number of stages, or functional form, nor does it include convergence analysis or the requested ablation. We will revise the methods section to include these elements: a linear ramp schedule for the plasticity parameter over a specified number of epochs, a brief convergence argument based on the joint optimization of loss and parameters, and an ablation comparing the proposed curriculum against abrupt transitions and no curriculum (showing divergence or loss of wrinkle persistence in the latter cases). revision: yes

  2. Referee: [Evaluation] The evaluation section claims the method 'outperforms existing methods on a variety of garments, body shapes, and body motions, according to a range of metrics,' yet the abstract provides no quantitative results, specific metrics for wrinkle persistence, or baseline comparisons. Without these, it is impossible to verify whether the persistent-wrinkles claim holds or reduces to visual inspection.

    Authors: The evaluation section of the full manuscript contains quantitative results using metrics that include a wrinkle persistence error (measuring deviation from plastic deformation targets), visual fidelity scores, and direct comparisons against prior self-supervised elastic baselines across multiple garments, body shapes, and motions. However, the abstract summarizes these claims without numbers or metric names, which limits immediate verifiability. We will revise the abstract to include key quantitative highlights (e.g., percentage improvements on persistence metrics) and ensure the evaluation section explicitly names the wrinkle-specific metrics and baselines. revision: partial

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: derivation remains self-contained

full rationale

The paper presents a novel loss that converts training into a moving energy minimization and a curriculum that gradually introduces plasticity parameters. Neither step reduces by construction to a fitted input or prior self-citation; the curriculum is introduced as an empirical training heuristic without any claim that it is mathematically forced by the loss definition itself. No equations are shown to be tautological, no uniqueness theorem is imported from the authors' prior work, and the central result (persistent wrinkles via self-supervised optimization) is not renamed from a known pattern. The derivation therefore stands on independent modeling choices rather than self-referential reduction.

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

Abstract-only view supplies no concrete free parameters, axioms, or invented entities; none can be extracted.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Self-supervised Garment Dynamics with Persistent Wrinkles." pith.science (2026). https://pith.science/paper/RFZOQDKY

@misc{pith2026260625065,
  author       = {Pith},
  title        = {Pith review of: Self-supervised Garment Dynamics with Persistent Wrinkles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RFZOQDKY}},
  note         = {Machine review of arXiv:2606.25065}
}
read the original abstract

Self-supervised neural garment simulation has become popular due to its computational efficiency, good visual realism, and no reliance on training data. However, existing methods greatly simplify the mechanical properties of fabrics, ignoring persistent wrinkles caused by plasticity. Although this simplification allows for modeling of purely elastic material and simple training via energy minimization, the lack of believable wrinkles adversely affects the visual realism. Therefore, we introduce the first self-supervised neural garment simulator that explicitly models persistent wrinkles. This is accomplished through a novel physics-inspired loss function, which turns learning into a moving energy minimization problem to mimic plasticity. However, this requires learning to use a changing loss function, which causes difficulties in training because the loss function changes during optimization. To this end, we propose a new physics-inspired curriculum learning scheme where the target material for learning gradually changes from pure elasticity to elasto-plasticity, allowing the loss function and the learnable parameters to jointly converge. Through a comprehensive evaluation, we show that for the first time, self-supervised learning models can generate natural persistent wrinkles, outperforming existing methods on a variety of garments, body shapes, and body motions, according to a range of metrics.

Figures

Figures reproduced from arXiv: 2606.25065 by the authors.

Figure 1
Figure 1. Given the bending waist motion (a), the physics-based cloth simulator (PBS) can simulate persistent wrinkles around the abdomen area (b) caused by the mate￾rial plasticity of fabrics. However, existing self-supervised cloth simulators (PBNS [5], SNUG [33], NCS [6], HOOD [14], and SENC [23]) naively model cloth as elastic ma￾terial and cannot simulate persistent wrinkles (d, e, f, g, h). Our novel self-supervised sim… view at source ↗
Figure 2
Figure 2. Elastic Net (E-Net): The input consists of a pose and its first-order time deriva￾tive. After passing through the encoder, the RB predicted by P-Net is used as a condi￾tional input to the decoder, which then predicts the final garment deformation S (1:T ) . Based on the predicted deformation, the RB is computed and subsequently used for P-Net training [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Plastic Net (P-Net): Hinge-level node and edge features at time step t are first embedded by MLP encoders. We then apply 3 message-passing steps, where edge fea￾tures are updated using an edge MLP fv→e(eij , vi, vj ) and node features are updated using a node MLP fe→v(vi, meanj (eij )). After decoding, the Rest Bending (RB) com￾puted by Eqs. (4) and (5) from the E-Net predicted deformation S (1:T ) as the training t… view at source ↗
Figures from the paper (18 more)
Figure 4
Figure 4. Figure 4: Visual comparison between our simulator and PBS across unseen bodies and motions: slim, and obese. Trained on the normal body, our simulator can generate wrinkles that are closely match PBS (a,b). Furthermore, it can generate plausible wrinkles on the slim and obese bo…
Figure 5
Figure 5. Figure 5: A comparison of wrinkle patterns across different garment types. (a), (c), and (e) present the PBS simulation results for the long-sleeve top, vest, and T-shirt, re￾spectively, whereas (b), (d), and (f) show the corresponding predictions generated by our model. ration.…
Figure 6
Figure 6. Figure 6: Our simulator allows users to tweak the fabric plastic properties. (a) By setting εy to 30◦ , the simulated garment, like paper, is very likely to form wrinkles after a bending motion. (b) After increasing εy to 60◦ , the garment forms fewer wrinkles. (c) As εy is set …
Figure 7
Figure 7. Figure 7: Convergence behavior under different alternating cycle settings [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 7
Figure 7. Figure 7: Convergence behavior under different alternating cycle settings. (a) (b) [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Training dynamics and long-term stability of our method. (a) Rest Bending (RB) and Predicted Rest Bending (PB) evolution across alternating training iterations. (b) Long-term autoregressive error over rollout frames. We additionally tested 10 randomly selected long mot…
Figure 1
Figure 1. Figure 1: Compare the persistent wrinkles formed by a waist bending motion. By mod￾eling cloth as elasto-plastic material, Physics-based Simulator (a) can simulate the persistent wrinkles in the belly area of the t-shirt. Our simulator is capable of closely mimicking the wrinkle…
Figure 2
Figure 2. Figure 2: Compare the persistent wrinkles formed on a t-shirt worn by a obese body. The results demonstrate our simulator (b) can constantly simulate persistent wrinkles which are close to Physics-based Simulator (PBS) (a) even if the body shape is varied. Whereas the garments s…
Figure 3
Figure 3. Figure 3: A comparison of wrinkle patterns across different garment types. (a), (c), and (e) present the PBS simulation results for the long-sleeve top, vest, and T-shirt, re￾spectively, whereas (b), (d), and (f) show the corresponding predictions generated by our model [PITH_F…
Figure 4
Figure 4. Figure 4: Compare the persistent wrinkles formed on a pair of trousers caused by squat￾ting. Our simulator (b) can closely simulate the persistent wrinkles as the Physics-based Simulator (PBS) (a). On the contrary, the trousers simulated by SNUG (d) and NCS (e) are nearly flat a…
Figure 5
Figure 5. Figure 5: Compare the persistent wrinkles formed on a vest caused by a bending waist motion. The belly area of the vest is bended greatly due to the motion (middle row). Only our simulator (b) can closely simulate the persistent wrinkles as Physics-based Simulator (PBS) (a) resu…
Figure 6
Figure 6. Figure 6: RB/PB evolution across alternating iterations. Temporal Dynamics of Plastic Bending In physics, (perfect) plasticity can be viewed as a dynamical system whose state is the rest bending (RB) angle. The evolution of RB depends on two factors: the current RB and the curre…
Figure 6
Figure 6. Figure 6: RB/PB evolution across alternating iterations [PITH_FULL_IMAGE:figures/full_fig_p025_6.png]
Figure 7
Figure 7. Figure 7: Ablation study on the number of message-passing steps in P-Net. The three columns correspond to 1-step, 2-step, and 3-step message passing. (a–c) Training curves of P-Net. The blue curves represent the training loss and the orange curves represent the validation loss. …
Figure 7
Figure 7. Figure 7: Ablation study on the number of message-passing steps in P-Net. The three columns correspond to 1-step, 2-step, and 3-step message passing. (a–c) Training curves of P-Net. The blue curves represent the training loss and the orange curves represent the validation loss. …
Figure 8
Figure 8. Figure 8: Visual comparison of different models for plastic bending dynamics. Compared with RNN (b) and LSTM (c), Transformer (d) and our message-passing GNN (e) can simulate more obvious wrinkles. denote the garment motion by {S(t) |t ∈ Z, t ∈ [1, T]}, where S (t) = {x (t) , x˙…
Figure 9
Figure 9. Figure 9: Elastic Net (E-Net): The input consists of a pose and its first-order time deriva￾tive. After passing through the encoder, the RB predicted by P-Net is used as a condi￾tional input to the decoder, which then predicts the final garment deformation S (1:T ) . Based on th…
Figure 10
Figure 10. Figure 10: Plastic Net (P-Net): Hinge-level node and edge features at time step t are first embedded by MLP encoders. We then apply 3 message-passing steps, where edge features are updated using an edge MLP fv→e(eij , vi, vj ) and node features are up￾dated using a node MLP fe→v…

Review history (3 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G.S., Davis, A., Dean, J., Devin, M., Ghemawat, S., Goodfellow, I., Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, L., Kudlur, M., Levenberg, J., Mané, D., Monga, R., Moore, S., Murray, D., Olah, C., Schuster, M., Shlens, J., Steiner, B., Sutskever, I., Talwar...

    work page 2015

  2. [2]

    Advanced Computing Center for the Arts and Design: ACCAD MoCap Dataset, https://accad.osu.edu/research/motion-lab/mocap-system-and-data

    work page

  3. [3]

    In: CG International’90: Com- puter Graphics Around the World, pp

    Aono, M.: A wrinkle propagation model for cloth. In: CG International’90: Com- puter Graphics Around the World, pp. 95–115. Springer (1990)

    work page 1990

  4. [4]

    In: Seminal Graphics Pa- pers: Pushing the Boundaries, Volume 2, pp

    Baraff, D., Witkin, A.: Large steps in cloth simulation. In: Seminal Graphics Pa- pers: Pushing the Boundaries, Volume 2, pp. 767–778 (2023)

    work page 2023

  5. [5]

    Soft Matter8(12), 3342–3347 (2012)

    Benusiglio, A., Mansard, V., Biance, A.L., Bocquet, L.: The anatomy of a crease, from folding to ironing. Soft Matter8(12), 3342–3347 (2012)

    work page 2012

  6. [6]

    ACM Trans

    Bertiche, H., Madadi, M., Escalera, S.: Pbns: physically based neural simulation for unsupervised garment pose space deformation. ACM Trans. Graph.40(6) (Dec 2021)

    work page 2021

  7. [7]

    ACM Transactions on Graphics (TOG)41(6), 1–14 (2022)

    Bertiche, H., Madadi, M., Escalera, S.: Neural cloth simulation. ACM Transactions on Graphics (TOG)41(6), 1–14 (2022)

    work page 2022

  8. [8]

    In: Proceedings - 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021

    Bertiche, H., Madadi, M., Tylson, E., Escalera, S.: Deepsd: Automatic deep skin- ning and pose space deformation for 3d garment animation. In: Proceedings - 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021. pp. 5451– 5460.ProceedingsoftheIEEEInternationalConferenceonComputerVision,IEEE (Institute of Electrical and Electronics Engineers...

    work page 2021

Show all 43 references
  1. [9]

    In: Proceedings of the 2003 ACM SIGGRAPH/Eurographics symposium on Computer animation

    Bridson, R., Marino, S., Fedkiw, R.: Simulation of clothing with folds and wrin- kles. In: Proceedings of the 2003 ACM SIGGRAPH/Eurographics symposium on Computer animation. pp. 28–36 (2003)

    2003

  2. [10]

    ACM Transactions on Graphics (TOG)33(6), 1–11 (2014)

    Cirio, G., Lopez-Moreno, J., Miraut, D., Otaduy, M.A.: Yarn-level simulation of woven cloth. ACM Transactions on Graphics (TOG)33(6), 1–11 (2014)

    2014

  3. [11]

    IEEE transactions on visualization and computer graphics 23(2), 1152–1162 (2016)

    Cirio, G., Lopez-Moreno, J., Otaduy, M.A.: Yarn-level cloth simulation with slid- ing persistent contacts. IEEE transactions on visualization and computer graphics 23(2), 1152–1162 (2016)

    2016

  4. [12]

    In: The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2024)

    Gong, D., Mao, N., Wang, H.: Bayesian differentiable physics for cloth digitaliza- tion. In: The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2024)

    2024

  5. [13]

    Gong, D., Yang, Y., Shao, T., Wang, H.: Cloth animation with time-dependent persistentwrinkles.In:ComputerGraphicsForum.p.e70031.WileyOnlineLibrary (2025)

    2025

  6. [14]

    In: The International Conference on Learning Repre- sentations (ICLR) (2022)

    Gong, D., Zhu, Z., Bulpitt, A.J., Wang, H.: Fine-grained differentiable physics: a yarn-level model for fabrics. In: The International Conference on Learning Repre- sentations (ICLR) (2022)

    2022

  7. [15]

    In: ACM SIGGRAPH 2024 conference papers

    Grigorev, A., Becherini, G., Black, M., Hilliges, O., Thomaszewski, B.: Contour- craft: Learning to resolve intersections in neural multi-garment simulations. In: ACM SIGGRAPH 2024 conference papers. pp. 1–10 (2024)

    2024

  8. [16]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Grigorev, A., Black, M.J., Hilliges, O.: Hood: Hierarchical graphs for generalized modelling of clothing dynamics. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 16965–16974 (2023) Self-supervised Garment Dynamics with Persistent Wrinkles 17

    2023

  9. [17]

    In: Pro- ceedings of the 2003 ACM SIGGRAPH/Eurographics symposium on Computer animation

    Grinspun, E., Hirani, A.N., Desbrun, M., Schröder, P.: Discrete shells. In: Pro- ceedings of the 2003 ACM SIGGRAPH/Eurographics symposium on Computer animation. pp. 62–67 (2003)

    2003

  10. [18]

    IEEE Transactions on Pattern Analysis and Machine Intelligence 44(1), 181–195 (2020)

    Gundogdu, E., Constantin, V., Parashar, S., Seifoddini, A., Dang, M., Salzmann, M., Fua, P.: Garnet++: Improving fast and accurate static 3d cloth draping by curvature loss. IEEE Transactions on Pattern Analysis and Machine Intelligence 44(1), 181–195 (2020)

    2020

  11. [19]

    In: Proceed- ings of the IEEE/CVF international conference on computer vision

    Gundogdu, E., Constantin, V., Seifoddini, A., Dang, M., Salzmann, M., Fua, P.: Garnet: A two-stream network for fast and accurate 3d cloth draping. In: Proceed- ings of the IEEE/CVF international conference on computer vision. pp. 8739–8748 (2019)

    2019

  12. [20]

    ACM Transactions on Graphics (TOG)27(3), 1–9 (2008)

    Kaldor, J.M., James, D.L., Marschner, S.: Simulating knitted cloth at the yarn level. ACM Transactions on Graphics (TOG)27(3), 1–9 (2008)

    2008

  13. [21]

    In: Proceedings of the European conference on computer vision (ECCV)

    Lahner, Z., Cremers, D., Tung, T.: Deepwrinkles: Accurate and realistic clothing modeling. In: Proceedings of the European conference on computer vision (ECCV). pp. 667–684 (2018)

    2018

  14. [22]

    In: Proceedings Com- puter Graphics International, 2004

    Larboulette, C., Cani, M.P.: Real-time dynamic wrinkles. In: Proceedings Com- puter Graphics International, 2004. pp. 522–525. IEEE (2004)

    2004

  15. [23]

    ACM Transactions on Graphics (TOG)42(1), 1–20 (2022)

    Li, Y., Du, T., Wu, K., Xu, J., Matusik, W.: Diffcloth: Differentiable cloth simu- lation with dry frictional contact. ACM Transactions on Graphics (TOG)42(1), 1–20 (2022)

    2022

  16. [24]

    Advances in neural information processing systems32(2019)

    Liang, J., Lin, M., Koltun, V.: Differentiable cloth simulation for inverse problems. Advances in neural information processing systems32(2019)

    2019

  17. [25]

    In: European Conference on Computer Vision

    Liao, Z., Wang, S., Komura, T.: Senc: Handling self-collision in neural cloth simu- lation. In: European Conference on Computer Vision. pp. 385–402. Springer (2024)

    2024

  18. [26]

    1–6 (2025)

    Liu, A., Hu, K., Mo, C.A., Li, C., Wang, Z.: Extended short- and long-range meshlearningforfastandgeneralisedgarmentsimulation.2025IEEEInternational Conference on Multimedia and Expo (ICME) pp. 1–6 (2025)

    2025

  19. [27]

    In: Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)

    Liu, A., Hu, K., Mo, C.A., Wu, Q., Kang, W., Wang, Z.: Pb4u-gnet: Resolution- adaptive garment simulation via propagation-before-update graph network. In: Proceedings of the AAAI Conference on Artificial Intelligence (AAAI). vol. 40, pp. 7060–7068 (2026)

    2026

  20. [28]

    ACM Transactions on Graphics (TOG)34(6), 1–16 (2015)

    Loper, M., Mahmood, N., Romero, J., Pons-Moll, G., Black, M.J.: Smpl: a skinned multi-person linear model. ACM Transactions on Graphics (TOG)34(6), 1–16 (2015)

    2015

  21. [29]

    ACM Transactions on Graphics (TOG)30(4), 1–8 (2011)

    Martin, S., Thomaszewski, B., Grinspun, E., Gross, M.: Example-based elastic materials. ACM Transactions on Graphics (TOG)30(4), 1–8 (2011)

    2011

  22. [30]

    ACM Transactions on Graphics (TOG)32(6), 1–10 (2013)

    Miguel, E., Tamstorf, R., Bradley, D., Schvartzman, S.C., Thomaszewski, B., Bickel, B., Matusik, W., Marschner, S., Otaduy, M.A.: Modeling and estimation of internal friction in cloth. ACM Transactions on Graphics (TOG)32(6), 1–10 (2013)

    2013

  23. [31]

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

    Narain, R., Pfaff, T., O’Brien, J.F.: Folding and crumpling adaptive sheets. ACM Transactions on Graphics (TOG)32(4), 1–8 (2013)

    2013

  24. [32]

    ACM Transactions on Graphics21(3), 291–294 (2002)

    O’Brien, J.F., Bargteil, A.W., Hodgins, J.K.: Graphical modeling and animation of ductile fracture. ACM Transactions on Graphics21(3), 291–294 (2002)

    2002

  25. [33]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Patel, C., Liao, Z., Pons-Moll, G.: Tailornet: Predicting clothing in 3d as a func- tion of human pose, shape and garment style. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 7365–7375 (2020)

    2020

  26. [34]

    In: Computer Graphics Forum

    Santesteban, I., Otaduy, M.A., Casas, D.: Learning-based animation of clothing for virtual try-on. In: Computer Graphics Forum. vol. 38, pp. 355–366. Wiley Online Library (2019) 18 X. Yang et al

    2019

  27. [35]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Santesteban, I., Otaduy, M.A., Casas, D.: Snug: Self-supervised neural dynamic garments. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 8140–8150 (2022)

    2022

  28. [36]

    Springer Nature (2022)

    Stuyck, T.: Cloth simulation for computer graphics. Springer Nature (2022)

    2022

  29. [37]

    In:Proceedingsofthe14thannualconferenceonComputergraphicsandinteractive techniques

    Terzopoulos, D., Platt, J., Barr, A., Fleischer, K.: Elastically deformable models. In:Proceedingsofthe14thannualconferenceonComputergraphicsandinteractive techniques. pp. 205–214 (1987)

    1987

  30. [38]

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

    Tiwari, L., Bhowmick, B.: Deepdraper: Fast and accurate 3d garment draping over a 3d human body. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 1416–1426 (2021)

    2021

  31. [39]

    ACM Transactions on Graphics (TOG)40(4), 1–14 (2021)

    Wang, H.: Gpu-based simulation of cloth wrinkles at submillimeter levels. ACM Transactions on Graphics (TOG)40(4), 1–14 (2021)

    2021

  32. [40]

    ACM Transactions on Graphics (TOG)29(4), 1–8 (2010)

    Wang, H., Hecht, F., Ramamoorthi, R., O’Brien, J.F.: Example-based wrinkle synthesis for clothing animation. ACM Transactions on Graphics (TOG)29(4), 1–8 (2010)

    2010

  33. [41]

    ACM transactions on graphics (TOG)30(4), 1–12 (2011)

    Wang, H., O’Brien, J.F., Ramamoorthi, R.: Data-driven elastic models for cloth: modeling and measurement. ACM transactions on graphics (TOG)30(4), 1–12 (2011)

    2011

  34. [42]

    ACM Transactions on Graphics (TOG)38(6), 1–12 (2019)

    Wang, T.Y., Shao, T., Fu, K., Mitra, N.J.: Learning an intrinsic garment space for interactive authoring of garment animation. ACM Transactions on Graphics (TOG)38(6), 1–12 (2019)

    2019

  35. [43]

    Wang, X., Chen, Y., Zhu, W.: A survey on curriculum learning. IEEE transactions on pattern analysis and machine intelligence44(9), 4555–4576 (2021) Self-supervised Garment Dynamics with Persistent Wrinkles Supplementary Material Xiaoyuan Yang1 , Deshan Gong3 , Taku Komura3 , a...

    2021

Pith tools

Reviewed June 30, 2026 · model on record in the stance chip above.