It is insufficiently powerful to compute multi-view images at a high enough frame or data rate to support high quality dynamic three-dimensional displays with the current computer software and hardware. According to the correspondence between lens position and elemental image, a real-time computer-generated content generation method for integral imaging based on cross perspective is presented. A GPU-driven rendering pipeline (GDRP) with geometry instancing and instance culling technique is designed, and the number of triangles processed in the rasterization stage is reduced greatly. Several virtual dynamic scenes including various materials and millions of vertices are used to verify the performance of the GDRP method in the experiment. Experimental results show that the frame rate can reach 60 fps with 3840 × 2160 resolution.

中文翻译：

---

基于GPU驱动的交叉透视渲染管线的实时计算机生成的积分图像

以足够高的帧或数据速率计算多视图图像以使用当前的计算机软件和硬件来支持高质量的动态三维显示，其功能不足。根据镜头位置与原始图像之间的对应关系，提出了一种基于交叉透视的实时计算机生成的积分成像内容生成方法。设计了具有几何实例化和实例剔除技术的GPU驱动渲染管线（GDRP），大大减少了在栅格化阶段处理的三角形数量。包含各种材质和数百万个顶点的几个虚拟动态场景用于验证实验中GDRP方法的性能。实验结果表明，在3840×2160分辨率下，帧速率可以达到60 fps。