摘要
为了充分利用图形处理器(GPU)的强大计算力和并行处理能力,并有效克服CPU/GPU间数据传输的瓶颈,提出了一种新的基于GPU的曲面自适应细分算法.通过采用细分模板(SP),在GPU的顶点处理器上将从CPU上传送来的控制网格进行求值细分.给出了自适应细分层次的判定,以及通过带裙边的SP来解决可能出现的裂缝问题.将该方法用于Catmull-Clark细分曲面和Loop细分曲面的求值显示,并推广应用到其他类型细分,和GPU上的其他着色器组合使用,对硬件要求很低,只需要能够支持顶点着色器的显卡.与CPU求值渲染、基于片段处理器求值渲染方法运行效率的对比分析,证明了该方法的高效性.
A new graphics processing units （GPU） based surface subdivision algorithm was presented to fully utilize GPU＇s powerful computation and parallel processing capability and overcome the data transferring bottleneck between CPU and GPU. The algorithm calculated the subdivision levels adaptively. Subdivision pattern （SP）was used to execute computation for the control net data transferred from CPU to GPU running on vertex processors. SP can be extended by using fringes to solve the watertight problem for adaptive subdivision. The algorithm was applied to Catmull-Clark scheme and Loop scheme and can be easily extended to other subdivision schemes with other GPU shaders. The algorithm has the minimum GPU requirement and can run on any GPU supporting vertex shaders. Experimental results show that the algorithm is superior to the CPU based and fragment processor based algorithms.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2008年第7期1145-1149,1217,共6页
Journal of Zhejiang University：Engineering Science
基金
国家"973"重点基础研究发展计划资助项目(2006CB303106)
博士学科点专项科研基金资助项目(20070335074)
国家"十一五"科技支撑计划资助项目(2006BAF01A45-05)
浙江省自然科学基金资助项目(Y107403)
杭州市科技计划资助项目(20062422B05)
关键词
图形处理器
细分曲面
自适应细分
顶点着色器
graphics processing units （GPU）
subdivision surfaces
adaptive subdivision
vertex shader
