Capturing and animating occluded cloth

This video demonstrates results from a new method for cloth motion capture that can produce detailed surface meshes for full garments. Traditionally, capturing complicated deforming surfaces such as cloth has been difficult due to substantial self-occlusion and folding. The method in this video overcomes these problems by preserving physical characteristics of the cloth and filling holes using a data-driven model. Examples include a dancing actor in a full garment, a heavily wrinkled sleeve, and a rag interacting with a solid object. It also demonstrates re-use applications of cloth capture data by binding skeletal mocap data to a data-driven skinning system.