基于GPU粒子系统的大规模雨雪场景实时模拟

基于粒子系统的雨雪模拟大幅提高了三维场景的真实感,但传统的基于中央处理(CPU)的粒子系统的渲染效率难以达到在大规模场景中进行雨雪渲染的要求.为此,提出了一种基于GPU的粒子系统来渲染雨雪场景的算法.该算法在视点前的一个固定区域内产生和绘制粒子,在顶点着色器中进行粒子属性的更新,在几何着色器中将粒子从点扩展为矩形,并对每一帧中的粒子的属性进行缓存处理,保证了粒子属性更新的连续性.此外,采用多幅雪花纹理与粒子随机组合,使雪花效果符合多样性和随机性.实验结果表明,该算法能在大规模场景中进行雨雪效果的实时渲染,并有较高的真实感.

Real-time simulation of rain and snow in Large-scale scene based on GPU particle system

Three-dimensional effect of rain and snow scenery by particle system is greatly enhanced.While,the traditional particle system based on CPU is hard to satisfy the real-time simulation,especially for large-scale scene.To solve this problem,a GPU based on particle system is put forward to simulate rain and snow.In this system,all particles is generated in perspective coordinate system in a cube area in front of the viewpoint,the attribute of particles is updated in vertex shader,and the particle points will be expanded to rectangles in geometry shader.Besides,the algorithm put several snow flake pictures to use,and each snow particle combine with one of the picture randomly,which makes snow correspond with diversity and randomness.The experimental results show that the algorithm can simulate rain and snow realistically in real-time in large-scale scene.