Synthetic Skies Using High Dynamic Range Images and Eigenskies

This paper presents a method to render synthetic sky images corresponding to given weather conditions. The method combines artificial neural networks and principal component analysis to associate the appearance of the sky with the state of a weather parameter vector. This association is then used to generate artificial sky images corresponding to a given weather parameter vector. The proposed method has important applications for example in flight simulators and in the game industry. The skies are represented by high-dynamic-range images which are able to store the dynamic properties of sky light. This representation can be used for global illumination in software packages such as Radiance to render scenes at arbitrary lighting conditions. The results show that, although the cloud details can not be represented by this method it is possible to distinguish between different weather states. CR Categories: I.3.7 [Computer Graphics]: Three-Dimensional Graphics and Realism—Color, shading and texture.