3D Animation Reconstruction Technology for Sports Skills that Integrates Motion Capture and Physical Simulation
Main Article Content
Abstract
Motion capture data lack physical constraints such as ground reaction forces and joint torques, leading to penetration and imbalance in the resulting three-dimensional animations. This paper proposes a constraint-adversarial physics-aware reconstruction method for sports skill animation. Physical priors are embedded into motion features through graph convolutional encoding with a differential dynamics loss function. A policy network is then trained via Proximal Policy Optimization, whose reward function combines a Dynamic Time Warping-based style term with angular momentum conservation and contact point stability terms to generate joint driving torques. The poses are updated through implicit integration, and penetration errors are corrected by a Signed Distance Field closed-loop module. Experimental results demonstrate that the method achieves a mean joint position error of 2.3 cm, a physics validity score of 0.92, an average reconstruction time of 18 ms per frame, and a naturalness score of 4.7. The approach preserves motion style while producing physically plausible animation.
Downloads
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
References
R. Rekik, S. Wuhrer, L. Hoyet, K. Zibrek, and A. H. Olivier, “A Survey on Realistic Virtual Human Animations: Definitions, Features and Evaluations,” Computer Graphics Forum, vol. 43, no. 2, pp. e15064:1–e15064:21, 2024. DOI: 10.1111/cgf.15064.
A. Kwiatkowski, E. Alvarado, V. Kalogeiton, C. K. Liu, J. Pettre, M. van de Panne, and M. P. Cani, “A Survey on Reinforcement Learning Methods in Character Animation,” Computer Graphics Forum, vol. 41, no. 2, pp. 613–639, 2022. DOI: 10.1111/cgf.14504.
T. J. Dobos, R. W. G. Bench, C. D. McKinnon, A. Brady, K. J. Boddy, M. W. R. Holmes, and M. W. L. Sonne, “Validation of pitchAI markerless motion capture using marker-based 3D motion capture,” Sports Biomechanics, vol. 24, no. 3, pp. 587–607, 2025. DOI: 10.1080/14763141.2022.2137425.
X. Wang, W. Zhang, C. Wang, Y. Gao, and M. Liu, “Dynamic Dense Graph Convolutional Network for Skeleton-Based Human Motion Prediction,” IEEE Transactions on Image Processing, vol. 33, pp. 1–15, 2024. DOI: 10.1109/TIP.2023.3334954.
D. T. Pham, Q. T. Pham, T. T. Nguyen, T. L. Le, and H. Vu, “A lightweight graph convolutional network for skeleton-based action recognition,” Multimedia Tools and Applications, vol. 82, no. 2, pp. 3055–3079, 2023. DOI: 10.1007/s11042-022-13298-w.
S. Starke, I. Mason, and T. Komura, “Deepphase: Periodic autoencoders for learning motion phase manifolds,” ACM Transactions on Graphics (ToG), vol. 41, no. 4, pp. 1–13, 2022. DOI: 10.1145/3528223.3530178.
J. Qin, Y. Zheng, and K. Zhou, “Motion In-Betweening via Two-Stage Transformers,” ACM Transactions on Graphics, vol. 41, no. 6, pp. 184:1– 184:16, 2022. DOI: 10.1145/3550454.3555454.
J. Won, D. Gopinath, and J. Hodgins, “Physics-based character controllers using conditional vaes,” ACM Transactions on Graphics (TOG), vol. 41, no. 4, pp. 1–12, 2022. DOI: 10.1145/3528223.3530067.
X. B. Peng, Y. Guo, L. Halper, et al., “Ase: Large-scale reusable adversarial skill embeddings for physically simulated characters,” ACM Transactions On Graphics (TOG), vol. 41, no. 4, pp. 1–17, 2022. DOI: 10.1145/3528223.3530110.
H. Zhang, Y. Yuan, V. Makoviychuk, Y. Guo, S. Fidler, X. B. Peng, and K. Fatahalian, “Learning Physically Simulated Tennis Skills from Broadcast Videos,” ACM Transactions on Graphics, vol. 42, no. 4, pp. 95:1–95:14, 2023. DOI: 10.1145/3592408.
B. Filtjens, B. Vanrumste, and P. Slaets, “Skeleton-Based Action Segmentation with Multi-Stage Spatial-Temporal Graph Convolutional Neural Networks,” IEEE Transactions on Emerging Topics in Computing, vol. 12, no. 1, pp. 202–212, 2024. DOI: 10.1109/TETC.2022.3230912.
Q. Men, H. P. H. Shum, E. S. L. Ho, and H. Leung, “GAN-Based Reactive Motion Synthesis with Class-Aware Discriminators for Human-Human Interaction,” Computers & Graphics, vol. 102, no. 1, pp. 634–645, 2022. DOI: 10.1016/j.cag.2021.09.014.
X. Zhang, Z. Chang, Q. Men, and H. P. H. Shum, “Physics-Based Motion Tracking of Contact-Rich Interacting Characters,” Computer Graphics Forum, vol. 45, no. 2, pp. e70336:1–e70336:9, 2026. DOI: 10.1111/cgf.70336.
J. Wang, T. Zhang, Q. Zhang, et al., “Implicit swept volume sdf: Enabling continuous collision-free trajectory generation for arbitrary shapes,” ACM Transactions on Graphics (TOG), vol. 43, no. 4, pp. 1–14, 2024. DOI: 10.1145/3658181.