Human Modeling And Skin Deformation In Data-Driven Realistic Human Animation

Computer animation is the combination product of computer graphics and art; it generates sequential virtual realistic figures in computer. Human animation is one of the most challenging research topics in computer animation area, because human bodies have more than 200 DOFs, very complex rules of motions, and the shapes of human bodies are personal and unique, besides human muscles would deform by the movement. The audience of computer animation are human beings themselves, and they are very sensitive to human animations like how human model moves and how the skin deforms during the movement. They can detect tiny difference between different human animations.The multiple layered construction method for deformable animated human model based on anatomical principles and motion capture data driven techniques are the new research in realistic human animation, which can produce vivid animations. Based on anatomical knowledge, we first build the standard template model including bone model and muscle model, and use 3D scanner to get real human skin model. We then use the motion capture date to drive these models to make human animations. The layered construction method can simulate the muscle deformation, mainly muscle stretches and bulges, during the motions, which will be reflected in the skin deformation to produce high realistic human animations.Generally the research on human animations focus on two parts: the human motion control and edit; the human model and skin deformation. This paper will focus on the second part, the main contributions of this dissertation include:1. Build a bone-muscle model as a template. The users do not need to worry about the shape of bones or muscles of a certain human skin model, they just need to set several local coordinate systems in the skin model. The bone-muscle model will be scaled and transformed to fit the skin model. The proposed technique makes it more convenient for the user to create and animate anatomically-based human bodies. It can relieve the burden of muscle modeling and adjusting process.2. A semi-automatically muscle parameter design method is proposed, which can find the action point for muscle resample process. It can save a lot of time while setting standard model parameter and also enhance the preciseness. The method needs curve reconstruction algorithm like Crust algorithm.3. For the template based anatomically modeling and data-driven human animation a vertex's coordinate system based free form deformation is proposed in this paper, the deformable object do not need to be contained in the deform-space.