So we've been told in the past that coloured spec maps are better..because....
Well, here is the thing, I was trying to write a tech docfor 3d palette discussing the use of spec maps (as part of a mapping discussion) - what they are, why we use them, but I could find NOTHING on the difference between black and white and coloured spec maps apart from vague theories with no proof.
Coloured ones are supposed to add a colouring to the reflection, and some people say that it works as a simple shader, increasing saturation around the terminator of the lighting, but I still can't find direct proof.
Another thing I was considering: Does it make that much difference? Sure, in high end high poly pre-rendered cinematics, but in a game engine will people even notice the subtle difference? Of you have 20 Orks/Space Marines/Hell Demons running at you, do you say "Oh, I like his rosy cheeks!" ?
Replies
I think the point you should try to make is that you should never feel that a specular HAS to be monochrome and that experimenting with colors will often times yeild very good results.
Yes because bad specmaps will make them look like 20 oversized action figures. Also, many games have scenes that don't have that much action, like cutscenes. Other games might involve no action at all (adventure games, for example). A game like Final Fantasy needs colored specmaps and SSS even though Unreal Tournament 2007 might not.
Also: http://mitglied.lycos.de/KDR_11k/pics/pimp/ingame_front.jpg. The "purple" parts are in fact black/grey, the purple comes from the specmap. The skin has a bluish specularity and looks less plasticy than standard Doom 3 characters.
For an iridescent beetle, say the diffuse colour is dark brown/black, you could give the spec a really saturated green colour, and that would give a really cool effect in game.
Likewise metallic paint for cars... generally easier to make certain materials look a lot more realistic with a coloured spec map.
For an iridescent beetle, say the diffuse colour is dark brown/black, you could give the spec a really saturated green colour, and that would give a really cool effect in game.
[/ QUOTE ]
I had a go at this exact thing a couple of months ago in doom 3.
Beetle Green Specular
Beetle Yellow Specular
Obviously now I look at it there's tonnes of areas for improvement and I'll go back and fix these at a later date I suspect.
Like on the metal/aged combo here
also on the red tube on the back. I used a yellow spec for that tube and it helped giving a ... different look
These maps can be funky
http://pioroberson.com/modelpics/ab_s.jpg
You might want to have a look at some doom3 textures for that, you can open the pak files with winzip and browse them easily. Some nice examples in there.
http://dynamic.gamespy.com/~polycount/ubb/Forum8/HTML/004130.html
Funny how the old board is still there, accessible but untouchable.
People always seem to use blue for skin specular, and when I ask why most poeple say, becuase umm..becuase...because he did it that way. Some people say that its because the earth has a blue skin, therefore things get tinted blue - yet thats assuming our models are outside on planet Earth on a sunny day.
So when I say I want technical explanation of why it is better, I CANNOT accept someone saying "Its better."
I can easily explain why a 24 bit image is better than an 8 bit. I can see the difference, but I can explain very easily the technical reason (there are more colours). I can explain why normal maps are better than grayscale bump maps, because I can see the difference and prove the technical difference. when writng a tech doc "Because" is not sufficient.
Let the reader's eyes do the explaining.
[ QUOTE ]
if I write a document thats says "Colour ones are better because I say so," then I have failed.
[/ QUOTE ]
How about "Colour ones are more useful because artists say so, you nerds!!"
Also keep in mind specular hotspots are not really a true to life thing to begin with but rather hack to simulate very small and varied reflections that appear because no surface is perfectly smooth. So describing why colored specular highlights look better than uncolored ones in scientific terms isn't easy.
http://www.androidblues.com/shadetut.html
^ An old Stahlberg skin shader tut that got me to think a bit about it.
Per, the opposite colour thing is what I was looking for, an explantion of why blue was used in skin, not just "use blue"
KDR, Again the stuff I was after. I mentioned in my very first about it emulating the saturation at the terminator, but I wanted some backup, and thats the stuff I was looking for.
I've been painting colour spec maps for a while, but was blindly following rules that might have been half truths - blue faces etc. I was painting my ears a much richer red in the spec map for SSS emulation, and it looked better, but like a I said, sometimes "It's better" Simply isn't enough of an explantion.
different types of objects have different physical qualities that may result in a different reflection effect. all metals, for example, have a colored spec hit. blue metal, blue spec hit, etc etc.
didn't everyone know that? am i missing something?
http://pioroberson.com/modelpics/info_kid.htm
http://pioroberson.com/modelpics/info_aquaboy.htm
Pior, what's the deal with the eye-specular you have in the Kid's fourth texture flat? I guess you're using that as an emissive pass on the eyes, in addition to the third-flat specular pass?
I'm thinking that you don't have control over specular width (to get those tight eye-reflections vs. the broad specular width for the skin), or the card sucks at specular... so you baked it in...?
I think Daz mentioned this a while ago, but for realtime eyes, instead of painting a sharp specular highlight onto the eyeball geometry itself, duplicate the front few polys of the eyeball, move them forward a bit, and use an opacity-mapped texture as a static highlight - so even if the eyeball rotates, the highlight will stay where it is.
I guess that's what that section is for on Pior's model.
Eric, yes I went for a painted-on specular for the eyeballs.
I tried it first with a 'correct' specular definition (high spec on the eye) but I disliked the results : at some point the hotspot, while looking good, was only to be seen on one eye or the other (model had no modelled eyes, they were simply here thanks to the nmap).
To avoid that I toned down the eye specs, added a little 'pushed' polygon shell in front of the eyes and gave that a prepainted specular spot. With a bit of selfillumination it faked a bright spec effect quite well and gave me a complete (even if hacked!) control over specspread.
Since the shell was slightly floating in front of the eye surface it gave the painted spec a parallax effect that faked 'reflection slide' as the model rotates.
That was inspired by a post Daz and/or Soul made a while back about the Bond models they were making.
I think it's really handy even if it goes 'against' the so-called theory about next gen chars
With the kid I wanted to stay with the same falloff setting for the whole model, but I think that in the case of a strong nextgen game engine one could define 'specular spread' on a permaterial basis thanks to a shader. Still need to play with that myself
[edit] Ha Morpheus beat me to it. You win a cookie Moppy!