50 likes | 397 Views
서울대 산학협동 프로젝트 과제 브리핑 Generating a Global Illumination Map on GPU. 2008/09/12 ㈜ 엔씨소프트 Team HON 김 철 웅. About Global Illumination. Local Illumination vs. Global Illumination
E N D
서울대 산학협동 프로젝트 과제 브리핑Generating a Global Illumination Map on GPU 2008/09/12㈜엔씨소프트Team HON김 철 웅
About Global Illumination • Local Illumination vs. Global Illumination • Local illumination only is enough when rendering an outdoor scene for a game since light sources are so strong that indirect lighting can almost be ignored. • When rendering an indoor scene, there are relatively weak light sources and the light rays emitted from them are much more scattered, reflected, refracted.
About GPGPU • Graphic processors are specialized for processing graphics pipeline which is not complicated as CPU’s. However, they can be utilized as advanced FPUs. • Moreover, a modern graphic card has more than two dozen processors with ultra high frequencies! • Now, why don’t you use GPU for calculating massive data set? • e.g) signal processing, neural networks, cryptography, pattern matching etc.
Milestones • M0. A crash course on computer graphics • - Basic Graphics Pipeline • - Programmable GPU • - General-Purpose computation on GPU(GPGPU) • - Global Illumination (GI) • M1. Generation of GI Map on GPU • - Design • - Selecting preferred GI algorithms • - Breaking down GI into smaller concepts • - List up computations which can be done by GPU • - Constructing up a general abstraction layer • - Implement/Debug • - Assessment of the result