基于深度剥离的屏幕空间环境遮挡算法

Screen-space ambient occlusion algorithm based on depth-peeling

针对原有屏幕空间环境遮挡算法存在的自遮挡距离误差问题,提出了一种基于深度剥离的屏幕空间环境遮挡算法(Depth-Peeling based Ambient Occlusion,DPAO).该算法通过深度剥离过程萃取了多深度层次的Z-Buffer,为遮挡计算提供了更精确的遮挡采样片段,同时利用深度剥离技术进行了修正,从而得到了更具有物理真实性地渲染结果.整个流程充分利用了现有GPU的SIMD特性并行实现.实验结果表明,该算法能够在高帧速率情况下为具有丰富高频细节的几何场景产生真实感强的柔和阴影,适用于视频游戏、科学仿真等对实时性要求较高的应用领域.

For solving the self-occlusion distance error problem in the previous screen-space ambient occlusion （SSAO） algorithms, a new screen-space depth-peeling based ambient occlusion （DPAO） algorithm was presented. The algorithm extracted more Z-Buffer with different depth level using the depth-peeling technique, and offered higher computational accuracy and physical reality for calculating the screen-space ambient occlusion. Furthermore, the algorithm was implemented in parallel with utilizing the SIMD characteristic of the GPU. The experimental results show that the present algorithm can generate more realistic and soft occlusion shadows for the geometry scene with high-frequency detail. It is suitable for such application fields with high real-time requirement as video game and scientific simulation.