For years, the users of visual systems, in both the military and commercial worlds, have made a plea to the developers to significantly lower the costs of those visual systems. Then, in a second breath, they have continued to demand high visual fidelity. Not surprisingly, the trend among visual system developers has been to provide more capabilities for higher cost. This paper describes a system design where a serious and concerted effort has been made to lower costs, while still maintaining the features most essential and effective for training tasks. During the highly selective process of determining system features, some capabilities, such as smooth shading, color blending and transparencies, were seriously questioned. Other capabilities, such as texture, scene detail management, resolution, dynamic coordinate systems, and reasonable image quality, remain high on the list. The paper describes how the essential features were incorporated in a highly cost-effective design with surprising flexibility and modularity. Painstaking efforts were made to optimize the hardware efficiency, making use of custom, semicustom, and commercial VLSI. These latest technologies have made possible a parallel processor architecture in the geometric processor-a departure from the traditional pipeline approach. The architecture provides system capabilities ranging from low-cost, night-only operation to high-resolution, daylight scenes rivaling the current high-end systems. All of this can be contained in a single card cage per channel. The challenge of the visual users has been met.