Lighting Design in Interreflective Environments Using Hopfield Neural Networks

Abstract

This paper proposes a method for calculating luminous intensity distributions of multiple light sources taking into account both direct and interreflected light when a desired luminance distribution in a space is given. This inverse lighting problem is useful for designing rooms and tunnels. In our approach, a luminance distribution is specified instead of an illuminance distribution, because it is closely connected to the appearance of rooms and is used for lighting design in a tunnel. To calculate the intensity distributions of light sources quickly and robustly, the property of Hopfield neural networks that their energy converges to a minimum is exploited. The proposed method should greatly facilitate lighting design when used with rendering techniques such as the radiosity method. Several examples including lighting design in a tunnel are shown to demonstrate the usefulness of the proposed method.