We propose a model to represent quasi-flat objects, such as ancient coins. These objects are flat surfaces, meaning their length and their width largely exceed their height, and feature a distinctive relief. This relief characterizes the object and its perception is directly influenced by the position of the object, the light direction and the viewer's direction. Our model is a single (non classic) image representation containing the underlying structural variations of the object. This model, that we call ``Multi-Light Energy Map'', is constructed out of several classic images taken with several illumination directions without computing the object's surface normals. Together with this information about the magnitude of the object variations, it is possible to compute the information about the angular part, that we call ``Multi-Light Orientation Map''. Using these two maps, it is possible to render an image of the object at any light azimuth, enabling the user to move the light sources during the visualization of the object. Moreover, these maps can be useful for the retrieval of ancient coins, either by robust image registration, or by object recognition using either contours or textures.