摘要
In virtual simulation application, it is often necessary to use Open GL to render large-scale 3D static scenes including urban architectures. Each scene unit generally has individual vertex data and texture. For large-scale data set, it is hard to render all scene units simultaneously. We need to render part of the scene separately, which is called the scene partition and culling. In general, we partition the whole scene into different units on the CPU. We present a scheme that optimize the GPU rendering pipeline to cull the large-scale static scene, which will reduce the CPU suspending time and take full advantage of GPU computing advantages to speed up the rendering efficiency.
In virtual simulation application, it is often necessary to use Open GL to render large-scale 3D static scenes including urban architectures. Each scene unit generally has individual vertex data and texture. For large-scale data set, it is hard to render all scene units simultaneously. We need to render part of the scene separately, which is called the scene partition and culling. In general, we partition the whole scene into different units on the CPU. We present a scheme that optimize the GPU rendering pipeline to cull the large-scale static scene, which will reduce the CPU suspending time and take full advantage of GPU computing advantages to speed up the rendering efficiency.
基金
Supported by National Key Technology Research and Development Program of China(2013BAK03B07)
National Natural Science Foundation of China(61232014,61421062,61173080,61472010)
National Marine public service project(201505014)
Equipment Development project(315050501)
