Optimal Combination of Ray Tracing and Rasterization (R squared)

目前，全球照明主要是用蒙特卡罗方法模拟，其中积分方程是通过生成随机路径逐步求解的。由于光线跟踪硬件已经存在了几年，这些方法已经成为部分实时的能力。另一方面，存在许多用于实时全局照明的基于快速光栅化的方法。因此，出现了两种方法的适当组合是否可以实现图像生成的加速的问题。因此，该项目的目的是找到光线跟踪和光栅化的最佳组合，这样在全局照明模拟的图像中不会出现可见的变化，但可以实现图像生成的显着加速。在蒙特卡罗模拟中，合适的光栅化技术将被选择性地沿着随机路径使用。此外，具有几何缓冲器的纯基于光栅化的实时照明应被光线跟踪器选择性地支持。

Today, global illumination is mostly simulated with Monte Carlo methods, where the integral equation is solved step by step by generation of random paths. Due to the ray-tracing hardware that has been available for some years, these methods have become partially real-time capable. On the other hand, many fast rasterization-based methods exist for real-time global illumination. Thus, the question arises whether a suitable combination of the two approaches can achieve an acceleration of the image generation. The aim of the project is therefore to find the optimal combination of ray tracing and rasterization, such that no visible changes appear in the image of a global illumination simulation, but a significant acceleration of the image generation is achieved. In the Monte Carlo simulation, suitable rasterization techniques are to be used selectively along a random path. In addition, a purely rasterization-based real-time illumination with a geometry buffer should be supported selectively by a ray tracer.