A Global Framework for Motion Capture

Human motion is so complex that model-based animation techniques still do not produce enough realistic trajectories. Consequently techniques based on real trajectories are commonly used in several application fields in order to animate human-like figures. Nevertheless the use of motion capture systems still remains difficult when the movements become complex (such as gymnastic figures) or are adapted to fit geometric constraints. We propose a software system to overcome several of these drawbacks and make it possible to directly apply captured trajectories to virtual actors. This system deals with the loss of data encountered in all optical systems, the various skeletons morphologies and the blending of several motions together in real time.