Real-time Shadow Rendering Method And GPU Acceleration

As a key technology to increase the realism of the virtual scene----the shadow of the real-time rendering technology, in 3D games, animation, film and television, three-dimensional geographic information systems, computer-aided design and manufacturing and many other fields was widely used. The effect of shadow played an important role in the generation of realistic graphics and determining the location of objects. it had very large effect for people to understand the impact of three-dimensional scene. However, in the realization of the traditional methods, either image space or object space algorithms, all the transformation operation calculation of graphics pipeline were completed by the CPU, so the performance of algorithms was greatly constrained. It could not guarantee the resulting shadows on the physical accuracy, as the same time, guarantee the achievement of real-time rendering results. On the basis of examining a number of papers at home and abroad, we had improved the volume shadow generation algorithm in this paper, combining with image space contour detection methods and using GPU Accelerated technology to deliver real-time shadow rendering. Major research and innovation is reflected in the following two aspects:We improved shadow volume algorithm, adding a two-dimensional array that save the depth test value and the state vector, then it will solve the traditional shadow volume algorithm, Z-PASS and the Z-FAIL problems, according to the scene and the viewpoint position of free choice of two kinds of methods, at the same time solve their own shortcoming of shadow volume algorithm. Combining with image space contour detection algorithm could be convenient and efficient to achieve a shadow volume rendering, and solve the problem shadow volume algorithm can not solve large and complex scene renderingWe studied the GPU acceleration model based on contour detection image space shadow rendering algorithm. In this paper, we used the pipeline GPU programming processor and processing characteristics of parallelism to accelerate the speed of shadow rendering, to reduce the burden of the CPU in shadow rendering. At the same time, we solved the real-time problem existed in the traditional shadow rendering algorithm. So it could strike a balance between the accuracy and efficiency of shadow rendering.