Historically, much of the time and expense of developing visual environment databases has occurred in the process of customizing three-dimensional features to fit properly on the terrain skin, and this interdependence has often imposed limitations in terrain fidelity and feature placement and density. A new Evans & Sutherland system performs this terrain/feature marriage in real-time with special feature-conforming processes which are implemented in the CIG hardware and which rely on a depth-buffer visual priority solution. This allows modelers to optimize the terrain model for maximum fidelity, and create and organize the feature overlay without regard for the topography of the underlying terrain, greatly simplifying feature design and placement. Modelers can work at much higher levels of abstraction, while generating visual environments which are more accurate and realistic.

This paper describes a new set of modeling strategies which convert high-level feature representations into displayable databases. Broad-brush feature descriptions such as DMA DFAD or 2851 can be rapidly transformed into compact data structures which create dense high-fidelity visual environments. Geo-Typical and Geo-Specific features can both be readily accommodated where mandated by mission requirements, and advanced hardware instancing modes allow features to be highly customized with each placement, achieving high compression of the feature database. The development process may be largely automated, and feature and terrain production can be performed in parallel, greatly reducing database development time and cost.