Abstract
Synthesize facial animation, including facial and head motion, and hair dynamics are consuming. They usually require a lot of human resources and time, people hard to produce animation in the past. Our system requires no complicated process, ordinary people can easily use it. Also, it is a great invention that we animate facial as well as hair dynamics. Our convenience and functionality make the users unable to resist it. We develop a text-driven real-time animation system. The users only provide us face and hair models as well as text, we can immediately synthesize facial motion and hair dynamics like actual reading. Realistic animation can be produced in real-time without requiring any tedious preparations. In order to faithfully synthesize facial animation, we produce all kind of user-specific facial motion by performing deformation transfer. By analyzing text information, we synthesize realistic facial motions by rule-based approach. As for hair animation, we apply skeleton-based techniques to simulate delicate hair dynamics. Despite a large amount of hair strands, we can still real-time animate by our proposed methods. Our results show that our system produces realistic animation with no training process. Even though the user's models are complicated, we can still simulate in real-time.