| With the more extensive applications of virtual human in virtual reality, entertainment, surgical operation simulations, real-time and realistic display of virtual human become more important. One of the challenges is how to get fast and realistic skin deformations. In the past 30 years, researchers have made great efforts on this problem and got a lot of achievements. But there are still many problems in this field.This dissertation has deeply studied the realistic surface-model virtual human skin deformations. The works and evolvements that author has done are introduced as follows:1) Implementation of seamless virtual human skin modeling and deformation based on cross-sectional contourIn order to gain real-time animations, many applications use rigid surface virtual human models. These models don't have skin deformations and always cause distortions in the vicinity of joints when be animated. In order to solve these problems, we have adapted cross-sectional contour deformation method, and implemented the seamless modeling of the former model. Further more, we have implemented fast and smooth skin deformations based on rigid body animation frame.2) Presentation and implementation of a new skin deformation method based on generalized cross-sectionCross-sectional contour model is approximately regular polygonal model and can't represent the details of the human skin. Therefore, We proposed a new skin deformation method based on generalized cross-section. In this method, all the vertexes in the arbitrary skin mesh are divided into groups according to given rule, each group is defined as a generalized cross-section. Then we can gain smooth skin deformations by managing the generalized cross-sections. This method can hold the details of the given skin mesh, and therefore obtain realistic deformations when virtual human is animated. Because it adopts group deformation rules, this approach can easily change the grouping rules and granularity to meet different demands of various applications.3) Presentation and implementation of new skin deformation method based on motion captureIt is very simple for generalized cross-section skin deformation method to set grouping and deformation rules by experiments. But on demand of some applications, this method can't get satisfactory results in the vicinal of the shoulder. It can't simulate the muscle deformation effect and can't avoid the distortion in the shoulder. Hence we designed a new skin deformation method based on motion capture.With motion capture devices,we can get real human motion and deformation data, then the real deformation characteristic can be learned by the machine learning technology. After... |
