Human figure synthesis and animation for virtual space teleconferencing

Human figure animation is a widely researched area with many applications. This paper addresses specific issues that deal with the synthesis, animation and environmental interaction of human figures within a virtual space teleconferencing system. A layered representation of the human figure is adopted. Skeletal posture is determined from magnetic sensors on the body, using heuristics and inverse kinematics. This paper describes the use of implicit function techniques in the synthesis and animation of a polymesh geometric skin over the skeletal structure. Implicit functions perform detection and handling of collisions with an optimal worst case time complexity that is linear in the number polymesh vertices. Body deformations resulting from auto-collisions are handled elegantly and homogeneously as part of the environment. Further, implicit functions generate precise collision contact surfaces and have the capability to model the physical characteristics of muscles in systems that employ force feedback. The real time implementation within a virtual space teleconferencing system, illustrates this new approach, coupling polymesh and implicit surface based modeling and animation techniques.