Human-like Character Animation of Maize Driven by Motion Capture Data

摘要

In this paper, a method based on motion retargeting for human-like character animation of maize plant driven by motion capture data is presented. Generally, plants have no articulated structures thus the character animation of plant object is a hard task. Firstly, the curve-skeleton of the 3D maize plant model is extracted according to its media-axis pattern. Secondly, the mapping relationship between the source human skeleton and the target character skeleton is constructed, and the virtual joints of the target character skeleton are located in terms of the human skeleton proportions. Provided with rich motion capture data, our method could convert motions between hierarchically and geometrically different characters and output the human-like animation of maize plant. The experimental results show that the proposed method is effective and robust. Finally, the main advantages and limitations are discussed.