Given a three-dimensional array containing values of a function of three variables, the algorithm presented here draws an approximation of the surface or surfaces where the function attains a specified value. This is done by contouring two-dimensional subsets of the three-dimensional array and suppressing the invisible parts of the contours. The union of all the contour line parts approximates the desired surface. The suppression of the invisible lines is the most complicated part of the algorithm and is done by forming a silhouette of the processed subset of the array and testing lines against this structure.

Many systems implement plotter device-independent computer graphics by having a system plot package which outputs a plotter-independent code (here called metacode) and having a translating driver for each plotter which uses this code as input. The translator for this code can often be run with greatest efficiency on the computer which hosts the plotter. In NCAR's configuration, various computers will drive different plotters, making a portable metacode translator a desirable tool. Constructing a metacode translator which can drive the simplest devices and yet provide the potential to use sophisticated plotter hardware features is a stimulating challenge. The design and implementation of such a translator are described.

The scientific computer user presents a unique challenge to the graphics system designer. If the graphics system is properly designed and implemented, computer graphics becomes a vital research tool for the scientific user. These users' desires and expectations play an important role in every phase of the design process.This paper describes the philosophy of the graphics system at the National Center for Atmospheric Research (NCAR) and its implementation. Details of algorithms are not presented; rather the purpose has been to describe a successful computer graphics system which may serve as a guide for other designers who desire to provide practical computer graphics. This graphics system operates under a batch-mode computer system without using interactive terminals.

A fast algorithm for eliminating hidden linies from perspective representations of single-valued functions of two variables is presented. Compared with the most widely used algorithm for this problem, the algorithm presented here is as versatile, but much faster. Compared with a recently published algorithm for this problem, the algorithm presented here is more powerful and as fast. Compared with more general algorithms, the one presented here is much less complicated, since this algorithm is tailored to this problem.