Accelerated Ray Tracing dengan Bounding Volume Hierarchy dan Compute Unified Device Architecture

Abstract

Ray tracing is a technique in computer graphics that produced image by traversing ray from eye to every pixels on screen, and then the intersection checks between ray and objects are done. Ray tracing can produce a photorealistic image, but needs a long time to do the computations. One of the ways to shorten the computation time is by reducing the number of intersection checks between ray and objects. One of the popular methods is Bounding Volume Hierarchy (BVH).

In this application, Compute Unified Device Architecture (CUDA) is also used beside BVH. The CUDA’s usage purpose is to reduce computation time by doing parallel computing in the GPU. Computations done in GPU are computing ray directions from eye to every pixels and color of every pixels.

The result of this application is more than 70% rendering time reduced when compared to ray tracing application with only BVH. The maximum reduction is 86%.