A Physically-Based Algorithm for Motion Mapping

Zhihua Chen, Lizhuang Ma, Yan Gao, Xiaomao Wu, Zhong Li

Abstract: A new approach is proposed to motion editing based on motion capture data, i.e., motion mapping. The key to our framework is the spacetime optimization method based on physical model. For reducing the complexity of optimization problem, we map the captured motion to a new character by using a simplified model. This model is a transformation tool and determined automatically by using the motion analysis technique. The motion mapping process includes three sub problems of motion retargeting. In order to further speed up the procedure for finding the numerical solutions, we describe an efficient method to find an appropriate initial solution for the optimization process. Results show that physically realistic animation can be created with our mapping algorithm.