In the past computer graphics efforts, several researchers have demonstrated that polygon models can be used to produce images of curved surfaces that appear to be smooth and accurate. However, the authors know of no attempt to appraise such imagery by using multiple human observation ratings.

This paper presents a new methodology for resolving visible surface images of solid models derived from Boolean combinations of volumetric building blocks. The algorithm introduced here is an extension of well-established scan-line hidden surface removal procedures, and it integrates knowledge of a Boolean construction tree in the surface resolution process. Several hidden surface coherence properties are discussed in terms of their possible exploitation in the intricate solid model visualization process. While many of the earlier coherence techniques depend on a polygon environment in which surfaces and volumes do not intersect, the Boolean process can not afford that luxury because it is inherently required to handle intersecting volumes and surfaces. Initial tests indicate that substantial performance improvements over previous methods can be achieved with the algorithm described in this paper, and that these improvements increase as model complexity increases.

To introduce rendered surface display technology into the production design environment, many CAD operations envision a single color video display device for download processing of selected model pictures. Creation of a single image from a typical industrial CAD model involving a large number of higher order curved surfaces will normally require a minimum of several minutes' delay for data acquisition and visible surface computation. This paper describes a method that extends the visible surface operation to give the designer an interactive tool with which he can more effectively comprehend model composition. Instead of defining data for a single image, an intermediate three-dimensional structured file is created. A tailored display program manipulates the structured data file allowing a user to interactively create surface rendered images. Program capabilities include:

A general purpose method for generating shadows using a polygonal coordinate data base is presented. The method is based on an object space polygon clipping hidden surface removal algorithm. Output from the program is in the same three-dimensional polygon format as the input. Thus, a shadowed data environment may be easily created and viewed from any observer position with no additional depth sorting time required for the hidden surface removal process. Shadows can also be cast by more than one light source. Since the shadows are generated in object space, the results can be used for both visual display and numerical analysis.

A polygon hidden surface and hidden line removal algorithm is presented. The algorithm recursively subdivides the image into polygon shaped windows until the depth order within the window is found. Accuracy of the input data is preserved.The approach is based on a two-dimensional polygon clipper which is sufficiently general to clip a concave polygon with holes to the borders of a concave polygon with holes.A major advantage of the algorithm is that the polygon form of the output is the same as the polygon form of the input. This allows entering previously calculated images to the system for further processing. Shadow casting may then be performed by first producing a hidden surface removed view from the vantage point of the light source and then resubmitting these tagged polygons for hidden surface removal from the position of the observer. Planar surface detail also becomes easy to represent without increasing the complexity of the hidden surface problem. Translucency is also possible.Calculation times are primarily related to the visible complexity of the final image, but can range from a linear to an exponential relationship with the number of input polygons depending on the particular environment portrayed. To avoid excessive computation time, the implementation uses a screen area subdivision preprocessor to create several windows, each containing a specified number of polygons. The hidden surface algorithm is applied to each of these windows separately. This technique avoids the difficulties of subdividing by screen area down to the screen resolution level while maintaining the advantages of the polygon area sort method.