SCALE: Modeling Clothed Humans with a Surface Codec of Articulated Local Elements

SCALE. Given a sequence of posed but minimally-clothed 3D bodies, SCALE predicts 798 articulated surface elements (patches, each visualized with a unique color) to dress the bodies with realistic clothing that moves and deforms naturally even in the presence of topological change. The resulting dense point sets have correspondence across different poses, as shown by the consistent patch colors. The result also includes predicted surface normals and texture, with which the point cloud can either be turned into a 3D mesh or directly rendered as realistic images using neural rendering techniques.