← ArchivePaper2006
Capturing and Animating Skin Deformation in Human Motion
Data-driven model of skin and muscle deformation extracted from human motion, enabling realistic soft-tissue dynamics on animated characters.
Abstract
This paper presents a data-driven technique for capturing and animating the dynamic surface motion of the human body, including bending, bulging, jiggling, and stretching, using a commercial optical motion capture system with approximately 350 small markers placed on the muscular and fleshy parts of the body. The sparse marker sample is supplemented with a detailed subject-specific polygonal model, and a local reference frame defined at each marker is used to clean noisy data by merging disconnected trajectories and filling occluded-marker holes via PCA. To animate the model, the marker motion is factored into rigid body motion of near-rigidly segmented parts plus a residual local deformation approximated first by a quadratic transformation and then resolved with radial basis function interpolation. The method is demonstrated on dynamic activities such as punching, jumping rope, and belly dancing, and results are compared to conventional motion capture and synchronized video.
How to read this
- Category
- Capture system + data-driven skin-deformation method
- Contributions
- A data-driven technique to capture and animate dynamic body-surface motion (bending, bulging, jiggling, stretching) using a commercial optical mocap system with about 350 markers on fleshy regions
- A pipeline that cleans noisy sparse markers (merging trajectories, filling occlusions via PCA) using a subject-specific polygonal model and per-marker local frames
- Factors marker motion into rigid motion of near-rigid segments plus residual local deformation, approximated by a quadratic transformation then resolved with radial basis function interpolation
- Context
- A data-driven alternative to physical flesh simulation, in the lineage of marker-based motion capture extended to capture soft-tissue surface dynamics rather than just skeletal motion.
- Correctness
- Demonstrated on dynamic activities (punching, jumping rope, belly dancing) with comparison to conventional mocap and synchronized video; fidelity is bounded by the sparse 350-marker sampling and the quadratic-plus-RBF deformation model, so it captures observed soft-tissue motion rather than predicting unobserved dynamics.
- Clarity
- Accessible and practically oriented; a first pass conveys the capture-and-model pipeline, a second pass clarifies the marker cleanup and the deformation decomposition.
- How to read it
- First pass for the capture setup and the rigid-plus-residual deformation model; second pass on the cleanup and RBF steps if you work with marker data or soft-tissue capture.
Builds on
Nothing in the archive, this is a starting point.
Built upon by
Related work
- Data-driven Modeling of Skin and Muscle Deformation 2008 / SIGGRAPH
- Data-Driven Physics for Human Soft Tissue Animation 2017 / SIGGRAPH
- NIMBLE: A Non-rigid Hand Model with Bones and Muscles 2022 / TOG
- OSSO: Obtaining Skeletal Shape from Outside 2022 / CVPR
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →