本文根据运动模糊产生的原理,实现了一种几何无关的,基于GPU的快速运动模糊模拟算法;提出了一种迭代机制,让GPU可编程绘制管线在后置处理过程中,能够近似地模拟累积缓冲区的工作方式.本文还提出了一种改进方案,本方案能利用深度缓存解决了多帧画面混合时混合结果错误的问题,而且为运动轨迹提供了更多的线索.实验结果表明,本文算法可以在降低较少渲染效率的情况下,较为真实地模拟出物体运动或者旋转时产生的模糊现象.
This paper proposed a new technology to fast approximate motion blur generation. According to the mechanism of motion blur generation, the algorithm achieved the goal with the iteration on GPU. The iteration measure simulated the blur implementation of accumulation buffer in the post-processing. In addition, an approach of development was illustrated and practiced as well. The improved algorithm provided more clues to help understanding of the trace of motion, and improved the reality of the render- ing results at the same time. The experiment showed that this motion blur algorithm finished rendering with low cost of efficiency and produced vivid motion blur effect.