kpicott@alias.UUCP (Socrates) (12/13/89)
Thanks to all who replied to my query about linear and area lights. In the area of linear lights, two papers on analytical solutions were found. The first, by John Amanatides and Pierre Poulin has been submitted to Eurographics '90 and I'll hopefully get a look at that soon. The second, "Shading Models for Point and Linear Sources", ACM TOGS, 4(2), April 1985, pp. 124-146. by T. Nishita, I. Okamura, E. Nakamae, proposes an analytic solution to the diffuse component, but only under certain circumstances. The latter unfortunately reduces to numerical integration in the majority of cases where spline surfaces are involved, although a method of optimization is given that reduces computation time for the numerical integration. This method would seem to be suited to lighting parallel and perpendicular to the illuminated surfaces. There was also a paper entitled "A Comprehensive Light-Source Description for Computer Graphics", IEEE CG&A, July 1984, by Channing P. Verbeck and Donald P. Greenberg that approximates both linear and area light sources as a series of point sources. This is a compromise to numerical integration, but is still computationally expensive. In summary, the analytical solution to linear sources exists and is calculable, at least for the diffuse component. The specular component exists, but direct calculation is almost expensive as numerical integration. As far as area light sources are concerned.. no analytical solutions were found. In fact, from the work examined I was left with the impression that even if the solution existed it would not be very useful from a light illumination point of view (ie non-radiosity). (Comments?) -- Kevin Picott aka Socrates aka kpicott%alias@csri.toronto.edu Alias Research Inc. R+D Toronto, Ontario... like, downtown 110 Richmond St. East M5C 1P1 (416) 362-9181 x336