Image data is generated from scene data, in which the scene data include object elements in world space. Each object element has a surface that may be subdivided into a mesh and a plurality of these objects have a surface that is created in response to data from a master shape in canonical space. The surface is created by an affine transformation of the master shape and the master shape has a known area. A corresponding area of the transformed surface is calculated with reference to the adjoint matrix and the affine transformation. With this information, the surface areas in world space are used to determine light emission characteristics for the scene. The technique is particularly attractive when rendering images in real-time for application within a virtual set.