Fast Hill Shading with Specular Reflection and Cast Shadows

Abstract

Modern computer graphics allow the rapid application of visualization techniques which previously would have required prohibitive amounts of labor. A good example is hill shading which, when done by hand can take months for a single map. Computer generated analytic hill shaded plots can be generated in minutes and this allows such luxuries as the ability to experiment with the direction of illumination in order to reveal particular features. It is clear that applying a shading algorithm to a surface can reveal features which are invisible when displayed using contouring or simple pseudocoloring with height alone (Yazdani, Christov, and Berenyi, 1987; Robertson and O'Callaghan, 1985). However, most automatic hill shading of geophysical data has been done using only the simplest shading model; namely Lambertian reflection which assumes that the surface diffuses incident light equally in all directions. This usually is justified both on the grounds that highlights and cast shadows can obscure features and for reasons of computational efficiency (Horn, 1982; Yoeli, 1965). Although it is true that aspects of the image can be obscured using a more sophisticated lighting model, it also is true that sophisticated lighting can reveal useful information. Specular reflection is highly sensitive to the orientation of a surface and if the source is placed appropriately this can make fine surface details visible to the eye. This technique is appropriate particularly in a system which allows the interactive manipulation of sources, something which is becoming possible with today's high-performance workstations. Cast shadows, on the other hand, are useful in revealing the gross topography of the surface. The surface appears relatively flat and featureless without them. Both specular reflection and cast shadows can be used to greater advantage when a CRT is the output device which, because it has a greater dynamic range than print media, is capable of accurately revealing detail in highlights and in cast shadows which may be lost in print reproduction. In this paper a surface and illumination model is presented together with a simple scanline based implementation. Pointers also are given to places where the algorithm can be extended in the interest of greater realism (at the cost of greater computational expense). The major original contribution is the efficient compution of cast shadows using a newly developed "shadowline" algorithm. Figure 1 shows an image of part of the geoid (Christou, Vanicek, and Ware, 1988) displayed using the algorithm which is described in detail here.

Department

Center for Coastal and Ocean Mapping

Publication Date

1989

Volume

15, Number 8

Journal Title

Computers and Geosciences

Pages

1327-1334

Publisher Place

New York, NY, USA

Publisher

Pergamon Press

Digital Object Identifier (DOI)

10.1016/0098-3004(89)90096-4

Document Type

Journal Article

Link to Full Text

Share

COinS