用GPU实现双三次Bezier面片逼近Catmull-Clark细分曲面  

Approximating Catmull-Clark subdivision surfaces with bicubic Bezier patches on GPU

作　　者：王丽珠[1] 刘伟[1] 徐李娜[1] 

机构地区：[1]广东工业大学材料与能源学院,广州510006

出　　处：《计算机应用》2010年第12期37-39,共3页journal of Computer Applications

基　　金：广东省自然科学基金资助项目(8451009001001237)

摘　　要：将双三次Bezier面片逼近Catmull-Clark细分曲面的方法利用图形处理器(GPU)高速计算来实现,首先在CPU上载入封闭的Catmull-Clark细分曲面基网格并构建基网格的内部表示,再利用GPU计算获得Bezier曲面的控制顶点,随后将这些控制顶点和预计算的UV参数方块在GPU上合并处理得到网格模型的顶点并输出到内存,最后通过DirectX10渲染绘制出模型。由此不仅可以满足模型高细分的要求,而且能够实现模型的快速显示。This paper proposed the scheme of approximating Catmull-Clark subdivision surfaces with bicubic Bezier patches on Graphics Processing Unit（GPU）.First,the basis mesh of Catmull-Clark subdivision surface was loaded and converted into internal representation on CPU.Second,the control points of bicubic Bezier patches were computed by using the information of internal representation on GPU.Then the control points of bicubic Bezier patches and the parameters of precomputed UV parametric squads were read by GPU and used to generate the control vertices of the resulting mesh.Lastly the resulting mesh was rendered by making use of DirectX10.By which,the higher quality of the object can be obtained and the speed of rendering can increase.

关 键 词：双三次Bezier曲面片 CATMULL-CLARK细分曲面 图形处理器 DIRECTX10 细分曲面 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]