一种基于视角剖分的可见性算法

A Visibility Algorithm Based on Subdivision of View Angle

文中提出了一种适用于在给定视点运动路径情况下,处理三维场景的可见性预处理算法。首先对所有可能的视线方向所对应的立方体进行细分,然后对分割后的平面沿着运动路径移动所形成的光束体进行可见性预处理,最后合并这些光束体得到对该运动路径的可能可见面集合(PVS)。由于能够有效的剔除被较大的物体遮挡的小物体,使得PVS中的面片数量远小于整个场景的面片数目,这样就可以避免绘制完全不可见的物体,从而提高场景的实时绘制的速度。

This paper presents a visibility preprocessing algorithm for 3D scene, which is suitable for the occasion that the moving path of viewpoint is given. We use a subdivision of the cube which is corresponding to every orientation, to process the visibility by means of the light volume, and to calculate a conservative potentially visible set for a given path. The proposed method could efficiently cull the small objects which are occluded by other large objects, so the number of potentially visible objects on each viewpoint are much less than the number of objects in the entire scene, and it can improve the speed of rendering scene for it avoids rendering the invisible objects.