一种基于GPU的地形顶点法向量并行计算方法

GPU-based parallel computing method for topographic vertex normal vectors

为实现三维地形场景实时光照和阴影,采用顶点周围四点高程和GPU(显卡)寄存器绑定方法,借助顶点缓冲区和顶点着色器实时计算顶点法向量,对比主流地形着色算法,实验验证了GPU法向量实时计算效率和显卡兼容性.研究结果表明:地形场景为百万级(1 048 576)三角形时,GPU顶点法向量计算方法具备较高渲染帧率(66帧/s)和较好显卡兼容性.研究结论初步突破了真实感地形需要Normal Map的限制,有利于网络条件下快速高效渲染大规模真实感地形.

In order to achieve the real-time lighting and shadow of three-dimensional terrain, with the help of vertex buffer and vertex shader, the method calculates normal information in parallel based on floating-point computation power of GPUs which binds height values of neighboring four vertexes to registers. The experiment proves the GPUs parallel computing capacity and graphics card compatibility by comparing the popular algorithm. Research results show that the method has a higher frame rate （66 frame/sec） and excellent hardware compatibility when the terrain scene contains one million （1 048 576） triangle. The conclusion of this study breaks through the limitation which is the three-dimensional terrain shading need Normal Map. It is beneficial to the highly efficient rendering for large scale terrain on Internet.