Image information is doubled/halved through an iterative process. At each level, the image takes a different form and, by extension, context.