Path Tracing Isn’t Slow, Computers Are: Speeding Up the Real-Time Path Tracer

  • Richard Mark Cartwright University of Derby

Abstract

Since the introduction of the rendering equation in 1986, there have been numerous path tracing approaches, including Monte Carlo, Light Tracing and Metropolis Light Transport. This aim of this study is to test the scalability of the GPU accelerated path tracing implementation by Lapere (2016a). This will be achieved by attempting to make the implementation faster to find a point where it would be usable in a real-time situation by testing the performance of the path bounce levels, resolution and Samples Per Pixel per frame. The efficiency and benefits of using HDR images as ambient light maps and the use of Bounding Volume Hierarchies are also explored in the study. The conclusion of this study found that in the chosen implementation the best solution for a balance of both performance and visual fidelity is a render with a resolution of 768x432 with 3 bounces and 1 Sample Per Pixel per frame.
Published
2017-06-11
How to Cite
CARTWRIGHT, Richard Mark. Path Tracing Isn’t Slow, Computers Are: Speeding Up the Real-Time Path Tracer. Discovery, Invention & Application, [S.l.], june 2017. Available at: <https://computing.derby.ac.uk/ojs/index.php/da/article/view/253>. Date accessed: 22 aug. 2019.
Issue
Section
Articles