A relational data base management system has proven to be an extremely valuable tool for implementing AUTOCOLOR, an interactive graphics application system. AUTOCOLOR generates realistic color synthetic images on a raster display which will permit designers to aesthetically evaluate automotive body shapes even before a clay model is built. Both exterior body surfaces and functional parts constructed with the GMSOLID solid modeling system can be displayed. Color coded scalar values such as stress, pressure, or curvature can also be displayed on the surface of an object. AUTOCOLOR is completely interactive, using menus and screen selection to define parameters and options, select a viewing orientation, or mix a color. The geometry, view, lighting, and image generation information for displaying an object are all stored in tables which can be accessed directly by programs or interactively using data base commands. The tabular organization has proven to be very efficient from the programming point of view. It offers the advantages of array referencing without the overhead of following a linked list or hierarchical structure. The availability of data base functions provides a powerful tool for the manipulation of application data structures. For example, sorting the edge table to generate a different view of an object can be done by a a single command. Data base commands can also perform utility functions like generating the mirror image of a part without writing special purpose programs. Files containing all tables for an object can be saved which makes it possible to create subsequent images very quickly without regenerating previously derived information. The use of a general purpose data base manager reduces the programming effort to implement a system, provides great flexibility, and can be accomplished without loss of execution efficiency. The data base manager supports many of the functions that would have to be provided anyway. These include: space management, creation of data, building a data structure, and location of specific data items.

The traditional point source light used for synthetic image generation provides no adjustments other than position, and this severely limits the effects that can be achieved. It can be difficult to highlight particular features of an object. This paper describes new lighting controls and techniques which can be used to optimize the lighting for synthetic images. These controls are based in part on observations of the lights used in the studio of a professional photographer. Controls include a light direction to aim a light at a particular area of an object, light concentration to simulate spotlights or floodlights, and flaps or cones to restrict the path of a light. Another control relates to the color of the light source and its effect on the perceived color of specular highlights. An object color can be blended with a color for the highlights to simulate different materials, paints, or lighting conditions. This can be accomplished dynamically while the image is displayed, using the video lookup table to approximate the specular color shift