21 July 2017 Accelerating physical rainbow model with CUDA
Author Affiliations +
Proceedings Volume 10420, Ninth International Conference on Digital Image Processing (ICDIP 2017); 104204C (2017) https://doi.org/10.1117/12.2281773
Event: Ninth International Conference on Digital Image Processing (ICDIP 2017), 2017, Hong Kong, China
Abstract
Currently Lorenz-Mie theory is the most accurate technique to simulate rainbows. However, it is still very attractive to use a model based on geometric optics to simulate rainbows, even methods based on geometric optics are not accurate enough. Since the droplets in the nature are non-spherical due to gravity and surface tension, Lorenz-Mie theory has difficult in handling non-spherical droplets, but they are easy for ray tracers. Sadeghi et al. develop a model based on geometric optics and do some extra work to match the prediction of Lorenz-Mie theory. However, the model takes much time on calculating to get sufficiently accurate phase function of droplets. In this paper, we firstly implement the model on PBRT, and then accelerate it with CUDA. The experiment results demonstrate that our acceleration algorithm greatly improves the speed of the model.
© (2017) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Jinsen Zhang, Changwen Zheng, "Accelerating physical rainbow model with CUDA", Proc. SPIE 10420, Ninth International Conference on Digital Image Processing (ICDIP 2017), 104204C (21 July 2017); doi: 10.1117/12.2281773; https://doi.org/10.1117/12.2281773
PROCEEDINGS
6 PAGES


SHARE
Back to Top