Research And Implementation On3D Real-time Rendering Optimization Based-on Instancing

With the innovation of3D rendering technology, the more realistic rendering isincreasingly demanded. Since the realism of the scene is higher, achieve smoothrendering is more difficult. Therefore rendering acceleration has become a hotspot anddifficulty for computer graphics research. With the continuous development ofcomputer hardware technology as well as the escalation of independent graphicsprocessing unit GPU, rendering acceleration technology is also changing. GPU‘s speedis doubling every year, to support increasingly complex computing. Therefore,rendering acceleration technology is developing in direction of how to more effectivelyuse the GPU for computing and storage.Scenes with a large number of similar models are frequently encountered in thereal-time rendering, such as jungle trees in3D games, a lot of people in a stand of abasketball game. This article is dedicated to the research of Instancing-based renderingacceleration for these situations, and develops a rendering acceleration system which isindependent of an upper layer rendering platform. The main content of this paper is asfollows:Starting from the basic rendering pipeline technology, researching how to locate abottleneck in rendering pipeline and optimization methods for different bottleneck.Analyzing the bottleneck in the rendering of the scene with a large number of similarmodels, researching the Instancing-based optimization strategy, designing renderingacceleration system which is independent of the upper layer rendering platform.Researching Instancing to render a large number of static objects in situation of a largenumber of trees in the forest. Researching Instancing to render a large number ofdynamic objects in situation of a large number of crowds.Experimenting algorithms andthe result shows the optimized Instancing have good effect on render acceleration inspecific scene. After the completion of the related work, the paper summarizes the experimentalresults and outlooks the use in practice.