基于GPU的几何体实例化研究

RESEARCH ON GEOMETRY INSTANCING BASED ON GPU

在视频游戏中大量物体需要渲染的情况下,如何提高渲染速度成为了一个越来越富有挑战性的工作。对基于GPU的几何体实例化方法提高渲染速度的原理进行了研究。在分析比较当前已知的几种实现方法的基础上,着重研究了其中的静态批次和顶点常量方法,并在OGRE平台上作了性能测试实验,对该方法在提高渲染速度方面的效果进行了对比性验证。

It is a growingly challenging task to increase the rendering speed, when there are a huge number of objects to be rendering in video games. In this paper, we did a research on the principle of how the GPU based geometry instancing improves the speed of rendering. The analytic comparison of several existing known implementation ways was made first, and then we placed importance of research on two of them： static batching and vertex constants hatching. The performance testing experiments were carried out on OGRE platform for the comparative verification of their effectiveness in increasing the rendering speed.