Research On Some Key Problems Of Computer Animation

As a branch of Computer graphics, animation arises with the rapid growth of computer hardware and graphic algorithms. It aims at generating a virtual world, where animator could fulfill their ideas with ease. Animation is required to complete three pieces of work including modeling, motion control, and rendering, among which modeling of complicated objects, controlling and maintaining the volume of objects during motion, and fast rendering of large-scale scene have become hotspots. This paper looks into these problems and makes improvements on existing methods.This paper firstly summarizes the present development and key technologies in animation, introducing the basic method in modeling, motion control and rendering, particularly on those existing results related to methods devised in this paper. Modeling is the first step of animation, and image-based modeling is developed to model objects with complicated shape. Photometric Stereo Method has proven its superiority in the field of image-based modeling, but its practicability is still in question and its theory needs improvment. Aiming at the hotspots in this field, this paper analyzes the method from the aspect of operational manner, performance trait, basic principle and circulation flow, exploring into the application of highlight model, optimization of robust modeling algorithm and fast search algorithm, which are adopt by the re-acceleration process in the filed of Photometric Stereo Method. Thereby we proposed a conjugate gradient approximation algorithm which is specially designed for T-S lighting model. A method that efficiently accomplish the work of elimination and substitution of non-dispersive pixel area using multiple pictures is also specified in this paper. Now we are able to help accelerate the pivotal step of "High-dimensional normal vector searching" with time complexity of logarithmic rate by making pyramid-like images.When topic turns to object deformation, maintaining the volume of objects is the primary problem to solve. Based on the theoretical characteristic of implicit surface, this article dwells on the process of implicit surface maintaining the volume of objects during motion. Combined with interval analysis method with the properties of metaball density function, a fast and accurate criterion is presented here. By this criterion, the goal of maintaining volume during motion is achieved by fast calculation of volume through recursive division.Rendering of large-scale scene is considered as the vital factor that improves thesense of reality in computer animation. Based on the auxiliary memory technique in real rendering, this paper presents a mainframe of memory cache and demand-oriented loading principle by adopting memory-mapped file and multi-thread technology. Then we propose an approach to schedule the data flow between main memory and auxiliary memory by utilizing memory-mapped file. And the implementation of priority-based asynchronous loading, data prefetching and dynamic configuration is also fully described. According to the hierarchy of cache memory in the architecture of GPU and CPU, an optimization method in pretreatment period of offline geometric data arrangement is proposed, further exploiting the potentials of display card in the way that prefetching and caching could be achieved automatically. Path planning is another crucial topic in animation. After studying popular path planning methods, we propose an improved Oct-tree based path planning method, and present the approach of scene preprocess, connected graph construction, path planning, path smoothing and adjustment of speed and visual angle of camera in detail.