The rapid advancement in capabilities of computer image generation (CIG) visual systems has resulted in increased application of such systems from take-off and landing training to full-mission simulation. These full-mission applications dictate a need for the creation of high-fidelity data bases covering large geographical areas on the order of multiple thousands of square miles. As a consequence, the manual techniques employed in the past to create small, airfield vicinity data bases are not practical for generation of very large data bases. Clearly, some type of automated data base generation technique is required. Current trends are aimed at utilizing the Defense Mapping Agency (DMA) digital data base as source data and applying to it a computer transformation in order to arrive at a real-time CIG data base. However, current limitations of the DMA data, CIG system constraints, and training utility of the end product limit the amount of automation possible in the data base generation process. This paper discusses the evolution of the CIG data base generation process from totally manual methods to current trends toward full automation. Practical limits of automation and potential future developments are examined.