Improved model of IBL sunlight simulation
Ildar Vagizovich Valiev, Alexey Gennadievich Voloboy, Vladimir Alexandrovich Galaktionov · 2008
This paper considers the use of high dynamic range images as a source of natural daylight illumination in virtual scenes containing objects with complex optical properties (such as automotive multilayer paints). Physically justified algorithms for calculating the luminance of an object point and recognition of the sun to replace it by separate light source are outlined. An analysis of the correctness of the available high dynamic range images is performed, and their drawbacks are discussed. Two algorithms that allow to compensate the drawbacks of the representation of the visible sun in the high dynamic range images and thus perform a more accurate illumination simulation are proposed and elaborated. The correction of sun representation is based on the sky models adopted by the International Commission on Illumination (CIE). The modified illumination simulation algorithms make it possible to produce more accurate images. The algorithms are semi- automatic and robust with respect to errors in high dynamic range images. These algorithms are used in our computer graphics system for generating realistic images.