Analytical sky simulation

The sky plays a prominent role in everyday life, and much research has been conducted to simulate the sky and all the phenomena in it.

We investigate, implement, and analyze two phenomenological skylight models. One of these is the well-known Preetham model1, based on the Perez luminance model to perform fast sky simulations under clear daylight. The other is the Hosek-Wilkie model2, a significant improvement of the Preetham model. Taking into account some additional phenomena that determine the color of the sky, this method still manages real-time performance.

The conclusion of our evaluation is that the Hosek-Wilkie model is indeed an improvement upon the Preetham model, although there are some drawbacks as well. By design - a more extensive formula is used for simulation - the computational performance is 30% lower than the Preetham model. Furthermore, the Hosek-Wilkie method generates an increase in brightness at lower solar elevations, which goes against physical correctness - in reality, brightness decreases as the sun gets lower. Finally, under certain conditions, the Hosek-Wilkie model gives wrong results, which is attributed to the fact that it is a phenomenological model. However, in previous research, the same problem was encountered in the Preetham model as well.

This project was implemented in C++ with OpenGL.

[1] A practical analytic model for daylight (SIGGRAPH 1999). AJ Preetham, P Shirley, B Smits.

[2] An analytic model for full spectral sky-dome radiance (SIGGRAPH 2012). L Hosek, A Wilkie.

This project was finished on the 29th of November, 2012
Associated paper