A light rig is generated based on the source image's luminance information. First, the image is divided into a specified number of "layers" of luminance. Each layer is turned into a matte in which lit regions are white and unlit regions are dark. The layer's lit areas are then turned into lights in your scene, based on their size (amount of luminance) and proximity to one another in the layer.
The source image must be either an .hdr or .pic image, and works best if it is a spherical map. It's also important that the image has a enough contrast between the highlights and lowlights.
Using an .hdr image produces better, more accurate lighting with a larger range of contrasts. For more information about High Dynamic Range images, see Using High Dynamic Range (HDR) Images [Texturing].
