Abstract
Recently, there have been a great amount of studies in generating effective animations through motion capture. Among them, the problem of motion data reuse, for example to adapt existing animation information to various facial models, is still under extensive investigation. In this paper, we extend muscle-based method for reusing existing facial animation data and synthesizing the facial expression with facial expression description language.In our method, the positions of the face features (eyes, eye brows, mouths, etc.) on source marker (attach to actor or actress) and target facial model are located manually. At the same time, the movement of face feature is captured via motion capture. We will record the relationship between marker (feature point) and facial muscles for computing the transferring ratio. Before applying the motion data onto facial model, the global motion and the local deformation information must be separated. Then transfer each motion vector from a source animation to a target model with possibly different geometric proportions and mesh structure.In this paper, we also present a script language call “Facial Expression Description Language” (FEDL) to synthesize facial expression. The FEDL contains two components, one is primitive data type, and the other is primitive operation. In this mechanism, we can control the appearance and emotion of facial expressions through writing FEDL script.