The Game Engine, 3d Face Modeling And Facial Animation And Realization

Highly realistic 3D facial animation technology has been more and more applied to film production, facial surgery and remote conference. By integrating with facial expressions and facial muscle movements during talking, more lifelike virtual scenes in video game are gained. On the other hand, facial animation is very different to skinned mesh technology in the traditional body animation, especially in facial modeling, expression modeling, texture creation and animation generation. For these differences, the experts have developed several relational methods, but yet the current researches are not insufficient, especially for 3D facial modeling and animation generation during the development of video game with game engine.This thesis systemically analyzes and summarizes the current technology in facial modeling and expression animation, then proposes and realizes a series method of facial modeling and expression animation for a graphics engine. In the system design, we use reasonable system architecture and processes to give users the greatest creative freedom. In the facial modeling, this thesis proposes a more simple and efficient algorithm based on space deformation algorithm. By using the controlling mesh made up of MPEG-4 FDP and its subdivided meshes, a smooth deformation space is generated. Then the deformation on controlling mesh will transfer to face mesh through this smooth deformation space. In expression modeling, by using MPEG-4 FAP and attenuation function based on distance, the displacement of feature points and non-feature points is computed, with that the expression modeling sub-system is implemented to support six basic facial expressions. In texture creation, a new method is proposed, which can overcome the shortcomings of traditional methods and draw texture on 3D models directly. By using a GPU accelerated hierarchical structure, the 3D visualization texture creation sub-system is implemented, and the demands of interactive, intuitive and real-time are met. In the generation of expression animation, an animation generation method based on expression snapshot is proposed, by which the compatibility with Pose Animation in OGRE is archived, and the output file can be used by engine directly. The results demonstrate that the algorithms in this thesis can meet the demands of real-time, interactive, quality and compatibility with engine in facial modeling and expression animation.Finally, we summarize our current work and propose the direction of improving and the future work.