The GPU-based Rendering Of Natural Scenes Optimization Technology Research

The three-dimensional rendering of the natural scene is Virtual Reality Game Entertainment, geographic information systems, and so on common research stage, with a wide range of applications. Complex natural scenes, including not only large-scale terrain, also includes a huge number of various vegetation, etc., coupled with high-resolution texture images, the whole scene rendering data is staggering. Recent years, GPU performance growth continues strong growth in CPU processing power between the infinite complexity of the scene and limited graphics hardware unit, still need to ride between them using appropriate models and algorithms to the graphics scholars bridge, the goal is that in addition to improving the speed of the scene displayed, but also have a real effect on the visual.In this paper, give full play to the characteristics of high-performance programmable GPU and graphics API’s latest performance possible graphics rendering to the GPU, reducing CPU load, with a system inside the CPU has more resources to do physics, artificial intelligence other calculations, in order to improve the efficiency of the entire system. Therefore, this study is the use of a programmable GPU natural scene rendering optimization, mainly in the following research results achieved:1、Combine traditional levels of detail technology and instancing technology, take advantage of the latest features of graphics Application Programming Interface, judged by traditional CPU to determine appropriate LOD improvement through programmable GPU single pass.2、The traditional view frustum culling general on the CPU, and this paper, using the geometry shader characteristics and transform feedback mechanism to achieve a multi-pass rendering technology-first pass GPU-based view frustum culling, and the second time and then draw that ultimately need to display instances.3、Based on the traditional hardware occlusion culling technology, in the depth buffer depth image of the inside structure level, as a the occlusion image Next Mipmap method to achieve a occlusion culling algorithm based on hierarchical occlusion map image. Since no such traditional methods as the mask database previously calculated, the new algorithm is more suitable in a dynamic scenario. Finally, through the implementation of an outdoor nature scene demonstration system, draw the contrast performance before and after the improvement of the algorithm. Experimental results show that these optimization algorithm to reduce the the scene drawing geometric complexity, improve the frame rate, CPU usage also declined. The demonstration system uses a simple and quick support OpenGL library——SFML to improve the efficiency of development.