子計畫一：用戶端多核心嵌入式系統三維圖型應用程式發展(2/3)

The development of 3D computer graphics has been following two separate paths, one along the improvement of rendering quality and one along the speedup of rendering. The former leads to impressive results such as those in commercial 3D animation feature movies, but demands huge computational resources. The latter leads to the booming of the computer gaming industry and graphics processor industry. However the rendering quality in real-time computer graphics, even with the help of dedicated graphics hardware still lags behind those in offline rendering. Therefore one of the hot topics in computer graphics is to develop real-time global illumination programs to achieve both high-quality rendering and real-time interactive rates. Another recent trend is the growing popularity of smart phones such as Apple iPhone and Google Android handsets. With the rich multimedia contents on the smart phones, the demand for 3D graphics also increases. The availability of cloud computing service on the smart phones also shifts the focus of the application programs from the typical office applications to the entertainment applications. Therefore we take a leap of faith to predict that the real-time global illumination algorithms will be at the core of the rendering function in the future smart phones. The goal of this project is to develop advanced computer graphics functions, such as environment lighting and indirect illumination. We will also investigate how to adjust the rendering quality to meet the requirement of low power consumption. Our algorithm is based on the ray tracing method, which is much more flexible than the traditional GPU-based rasterization method in terms of adaptive rendering quality. We will also work with subprojects C and D to pursue heterogeneous computing on both multicore CPU and GPU using OpenCL. Another interesting topic to be investigated in this project is to utilize the cloud computing infrastructure, and to adjust the rendering quality based on the available network bandwidth.