Re: [Algorithms] reflectance + brdfs = trouble?

[Chris H. <ch...@d6...>]

>Well that was one of the main shortcuts Kautz and co made in the paper they
>presented, wasn't it? IIRC, They had a BDRF that had like six inputs, they
>dropped two by making it invariant across the surface, leaving four (light
>and view direction wrt surface normal), which then allowed for the equation
>two be separated into two lookup tables (textures) and two input terms.

Right.  They sort of ignored the texturing question after that, though.  If you've got a fully spacially varying BRDF then I guess you can work all sorts of stuff in there (scuffs, dirt, etc.) that might not be considered inherent albedo color (read: texture).  Still, it'd settle for a good way to know whether to + or * the albedo texture in after the BRDF.  :)

> Remember that you can have even more terms too, as the surface may be
>translucent and may scatter light internally. Skin and ice are good
>examples. If you've ever been on a frozen lake before, you've seen a good
>example of the latter (Canadian guy provides a sample :-) )

Debevec's example at the HCTS was skin, as you'll remember (although Canadian ice is a fine example, eh?).  It's got internal reflections and whatnot...think earlobe with a light behind it.  I think he said the next step up in lighting studliness was the BSSTF or something like that.  You end up with more-Ds because now light can come from any direction, be transmitting out anywhere, and you can look at it from anywhere.  Or something like that. :)

Chris
Game Technology Seminars, about a week away!
Physics & Graphics
www.techsem.com