Image-based lighting design

Abstract

In this paper, we present a lighting design framework for neardiffuse real objects, starting from a set of prerecorded photographs of an object under various lighting conditions. Light sources are placed at fixed positions around an object, for which the intensities are to be determined. Using existing digital imaging software, the lighting designer paints the desired illumination distribution on a photograph of an object. This painted-on illumination is used to determine the light intensities which produce a shading of the real object matching the desired illumination distribution. Certain areas in the painted image can be favored among others by weighting their importance in the optimization algorithm.