Path: snews.apol.com.tw!news.ht.net.tw!/home/news!feeder.seed.net.tw!ctu-gate!news.nctu.edu.tw!newsfeed.berkeley.edu!ucberkeley!howland.erols.net!portc.blue.aol.com.MISMATCH!portc01.blue.aol.com!newsstand.cit.cornell.edu!not-for-mail From: westin*nospam@graphics.cornell.edu (Stephen H. Westin) Newsgroups: comp.graphics.rendering.renderman Subject: Re: radiance renderings Date: 31 Aug 2001 10:21:08 -0400 Organization: Cornell University Lines: 74 Sender: shw5@cornell.invalid (on diesel.graphics.cornell.edu) Message-ID: References: <9mm93b$36kac$1@ID-75448.news.dfncis.de> <9mnrl4$35hkn$1@ID-75448.news.dfncis.de> NNTP-Posting-Host: diesel.graphics.cornell.edu X-Trace: news01.cit.cornell.edu 999267670 5663 128.84.247.211 (31 Aug 2001 14:21:10 GMT) X-Complaints-To: usenet@news01.cit.cornell.edu NNTP-Posting-Date: 31 Aug 2001 14:21:10 GMT X-Newsreader: Gnus v5.7/Emacs 20.6 Xref: snews.apol.com.tw comp.graphics.rendering.renderman:1985 "simon crowder" writes: > "simon crowder" wrote: > > > > I want to do some radiance renderings of architectural spaces. > > Can you recommend renderers which have good and fast > > global illumination? > > I have tried using BMRT but the results are very poor, do you > > know any good tutorial for rendering with radiance in bmrt? > > I meant radiosity not radiance, but thinking about it isn't that > the same? What is the difference between irradiance and > radiance? If you really want to know, point your browser to for a sort of "Cliff's Notes" of global illumination. Briefly, radiance is pretty much what a pixel expresses. Technically, that's energy flux per unit area ber unit solid angle. Irradiance is slightly different: it s the energy flux per unit area. The way I think of it is this. Imagine taking a sheet of film and uncovering it for some period of time. When you develop it, you will have a record of irradiance over the surface of the film: the total energy reaching each point, regardless of the direction whence it came. Now take another sheet of film and load it into a camera. Expose it similarly in the same environment, facing the same direction. When you develop that, it will be a record of radiance: energy from specific directions. Radiant exitance, or radiosity, is like irradiance, but in the opposite direction. Instead of the total energy received at a point, it's the total energy leaving that point. If your surface is Lambertian (constant radiance coming off: no gloss), the radiosity completely describes the light leaving a point of a surface. For more complicated surface characteristics (e.g. gloss), it's insufficient. An important distinction is that irradiance and radiant exitance basically obey the inverse-square law you learned long ago, but radiance doesn't. It is, in principle, constant with distance. This is why you don't have to squint when you move closer to the TV set; since your eyes are basically picking up radiance, things don't get brighter when you get closer. All this gets confused because of computer graphics history. In 1984, a paper from Cornell introduced the "radiosity" method to computer graphics. This is a traditional method (dating from at least the '30s) in heat transfer. It assumes Lambertian surfaces (so radiosity at each point completely describes what's goind on) and divides the environment into a bunch of discrete elements. This gives a big matrix equation that can, in principle, be solved to find the total energy balance (of heat or light) in an environment. This was, I think, the first method to compute diffuse indirect illumination (aka. global illumination) in computer graphics. In 1987, Greg Ward introduced the technique now known as "irradiance caching", an alternative way to make indirect illumination calculations practical. For better or worse, he named his renderer "RADIANCE". It's freely available and used fairly widely in illumination engineering (Greg was working for the Department of Energy at the time). I think some folks refer to the irradiance caching technique as "radiance", in homage (or at least reference) to Greg's work. So what you're trying to compute is diffuse global illumination; there are several approaches (even within BMRT) to compute this. -- -Stephen H. Westin Any information or opinions in this message are mine: they do not represent the position of Cornell University or any of its sponsors.