Newsgroups: comp.graphics From: mcdonald@aries.scs.uiuc.edu (J. D. McDonald) Subject: Re: Raytracing transparent objects. Date: Mon, 22 Feb 1993 16:40:25 GMT Organization: UIUC SCS -------------------------------------- [Another poster asks:] The problem: \ \ \ / <-reflection \ / \ / -------------------- \ \ \ <-refraction How much of the incoming ray is reflected, and how much is refracted, according to the angle, the intersecting ray hits the transparent object? I heard for raytracing this is only a constant. (Let's say, 20% are reflected, the rest goes through the transparent object.) But whats the solution to this problem in the real world? Any physicians out there? I assume (and every physics-book says the same), that the portions of the reflected/refracted ray depend on the angle (the smaller the angle, the smaller is the refracted portion, and vice versa), but in *what* way? Linear to the angle? Sinus? Arcussinus? Arcuscosecans hyperbolicus? :) Wolfram For light polarized in the plane of incidence the fraction R reflected is 2 sin (I - I') R = ----2------------ sin (I + I') and for that perpendicular to the plane of incidence, 2 tan (I - I') R = ----2---------- tan (I + I') where I is angle of incidence and I' angle of refraction. Of course sin(I') = N sin(I) / N' For glass N' = 1.523 versus air, here's a table: I (deg) R(%) parallelR(%) perp. 0 4 4 10 4 4 20 4 4 30 7 3 40 9 2 50 12 0.5 60 20 0.5 70 33 7 80 60 25 90 100 100 Ref: Smith, Modern Optical Engineering, McGraw-Hill , pp 167-168 For getting the appearance of a complicated scene right, you can't neglect polarization. You could do so only if all the angles of incidence are 40 degrees or less, or if there were no sequential reflection. Doug McDonald