Talk:ReflectanceFunctions

From Odwiki

Hi all, Should I start splitting this stuff up into separate pages? --mario 22:08, 12 Dec 2004 (PST)

---

Nope, I like it as it is -- Have you found the mozex plug-in? It makes editing pages soooo much easier --mark 11:31, 13 Dec 2004 (EST)

---

Ooooo... I just did a google on it. Very nice! I'll give it a try. Thanks! --mario 10:56, 13 Dec 2004 (PST)

---

Hmm, I can only find one for firefox 0.8, not 1.0. Anyone have it working for Firefox 1.0? --jason

---

We're using 0.9 here at work (there is a version available for 0.9 from the Firefox Extensions page). I just downloaded it and tried it (I'm writing this in VIM, yay!), and it seems to work well. Maybe the 0.9 version will work in 1.0? -- mario

---

I had to do some wierd hacking in the .mozilla directory to get it working properly, but boy, is it ever worth it... In the ~/.mozilla/firefox/*.default/prefs.js file, add something like:

   user_pref("mozex.command.textarea", "/home/username/bin/mozedit %t");

Where the script mozedit is something like:

   gnome-terminal --window-with-profile=Default --disable-factory -x vim $1

I can't remember whether there were any other hacks I had to do though. Sorry. --mark Dec 13 17:30:10 EST

[edit] Fresnel Question

Something's bugging me about Fresnel (damn, that thing is an endless source of head-scratching! . I never bothered implementing the full function before (never had a need), but now that I did (for completeness), it strikes me that the "lobe" that develops toward grazing angles as η heads off to infinity (and intensity at normal incidence approaches 1) is too pronounced -- i.e: it dips down too far. I have seen the shape before in graphs from different papers (I'll try to find some references), so I'm pretty sure it's supposed to be there, but I don't recall it being so big.

So I've double- and triple-checked the formula (including two full derivations) in case I had screwed up somewhere, but everything seems to check out... does anybody see something wrong in there? (I've been looking at this stuff so long that I'm afraid I might have missed something obvious).

Anyone else feel there's something wrong with that graph?

TIA, --mario 18:00, 5 Jan 2005 (PST)

[edit] RE: Fresnel Question

Huh!... fascinating stuff... I think I just found out what was happening.

The plots I remembered seeing were for conductors (metals) not dielectrics (glass). In the case of conductors we *have* to use a complex index of refraction for the second medium (η₂), which means we need to adjust the formula (which is the average of of the reflection amplitudes for s-polarized and p-polarized light) to account for this. The index of refraction for the second medium is then given by  η₂ − ik , where k is interpreted as (negative) absorption, with the effect that as k increases, less light is transmitted (and more reflected). -- in reality, the formula I show in the notes is just the special case where k = 0.

The trick is coming up with a complex-numbered version of the unpolarized formula given in all the CG-related references (the one I show in the notes), but I think I have enough references in front of me to cobble one together. The cool thing is not so much the fun of putting it together, but that after playing with what I've come up with so far, it seems like it could provide a wonderfully intuitive way to dial up the "metal-ness" of a surface just by sliding k (i.e: take a base ior, like say 1.3, and then go from plastic/glass to metal with one slider).

BTW: During my search for this stuff, I chanced upon a "way kewl" site with a really good interactive Java app showing all the aspects of the Fresnel equations in all their graphical glory (I'm a visual guy, what can I say). Check it out at the University of Barcelona.

Uh oh... suddenly I feel like I've been having a conversation with myself... that can't be a good sign... I'm finally starting to loose it... :(

--mario 14:53, 6 Jan 2005 (PST)

I've been reading but then I've nothing to contribute. -Edward

Hehehe... Oh good; guess I'll hang on to my sanity a little longer then
--mario 19:48, 6 Jan 2005 (PST)

---

I just discovered this Wiki and it rocks! Particularly this page is very nice. I suggest to web-in/add the info from this posting from mario.

--Moritz

---

Hi Moritz,
I'm glad you like it
The info that you point to in that post is already covered in these notes under the heading "The Lambertian Effect" -- and is explained in quite a bit more detail. The PI thing I covered much later under "Normalization And The Enigmatic PI" because it can really only be explained by looking at the BRDF as a continuous function (and integrating over the hemisphere).

Cheers!

--mario 07:35, 17 Jan 2005 (PST)