SSS map? What should it look like?
wetcircuit
Posts: 0
What are you using for an SSS map in people? I've only seen one (came with a texture at Rendo) and it was basically a big blob of hand-painted gauzian blur....
But what *should* be in an SSS map? Veins? Collagen? Sweat pores?
What I usually see is sort of a blob-y glow that wipes out highlights and detail (but does give some "warmth").... For me it doesn't seem worth the extra render time for something so basic, and I'm not really thinking it adds a lot of realism which is the whole point...
So what should we be seeing? How detailed should an SSS map be, considering it is a form of light scattering under a translucent surface...?
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Right.
What you really need is a map that illustrates those parts of the body that allow for the scattering effect. Like the upper, outer ear and outer portion of the nostrils, the part of the fingers outside the scope of solid bone, etc.,
Veins would be better displayed in bump or displacement, IMHO.
Anywhere where light will scatter and cause a glow when shone directly at or glancing past an area should be brighter on the scale - and the more opaque should be darker. True opaque should just remain black.
Homo Erectus also includes SSS maps
Also, as the skin thickens in those parts, so should the map get darker. Like the thickest skin of the nostril should be darker than where it is at its thinnest - as with the various shapes in the ear, etc.,
I am now holding my hand over the end of my spot lamp. My fingers show a good scattering all the way through, but a bit less towards the center - but the scatter makes it to the center from each side. So the map there should be lighter toward the edges. The main body of the hand had no scattering apparent at all - so that should be black.
Even though there are parts where scattering goes all the way through, there needs to be some sort of gradient as the parts get thicker (darker) - as long as it never reaches true blackness, the render engine will allow something to happen. We've all seen how effective the highlight channel can be by being extremely close to black, but not quite. The same will be true with the translucency of an SSS.
Is it worth the extra render time? Maybe for a very special moment in animation, or going for broke on a still. For most purposes, however, I think not.
If such an effect is wanted, however, one could use a good SSS map in the glow channel and cheat the effect. I learned that from Evil Producer! ;)
Here is a Maya technique.
In the first video, he is in Photoshop working with different layers for the skin (supplement for the Maya SSS shader).
The second video is in Maya, it shows the renders and he uses nodes (would have to use mixers in Carrara).
Not entirely translatable to Carrara, but I found the instruction clear and may give insight to one approach.
The first one is long, so grab some herbal tea.:)
(Start)Maya SSS shader tutorial 3-1
http://www.youtube.com/watch?v=wc4F-OL4y3w
(results) Maya SSS shader tutorial 3-2
http://www.youtube.com/watch?v=1_gjEs9aVtQ
what you are describing... it can't be accomplished with a map - at least not in Carrara. The map would not change according to the angle of light...
Turn you hand to the side and splay your fingers... If you had a map that makes the sides of each finger more SSS, but paints a dark stripe down the length of the bones... from a side angle that painted map would be wrong.... It would be like a halloween costume of a painted skeleton on a catsuit... there is no dimensionality to it, it's all on the surface....
I tried to skip to the second video in the Maya tut... It made no sense, so much for cheating... now I have to watch part 1 lol.
It would sort of make sense if the SSS map is a kind of depth map... the depth of fatty tissue directly under that point in the texture which scatters the light... So maybe it *should* just be a gauzian blob smear? at any point in the texture you are basically measuring the depth to the really solid stuff, so the ear for example would all pretty much be the same map level, because the geometry of the figure would be what gets calculated during the SSS process...
Hmmm.
True.
The entire effect is all a play on light and how it reaches the imaging structure within the eye.
Where you'll need subsurface scattering is in those intense moments when the light is just right. We don't walk around as glowing objects.
In Carrara, we've been theorizing the idea of using the SSS map in the translucency channel while having the SSS channel active. It is explained, I believe in the manual, that the SSS channel takes control of translucency when active. What I'm not sure about is whether or not we can use the translucency channel to override SSS - or at the very least, having the translucency setting in SSS control the strength of the Translucency channel... whew. Something on that order. EP will substitute the entire process, while saving an enormous hit to render times, by utilizing the glow channel. He has shown similar results, by testing the effects with the rest of the scene settings being equal. The Subsurface Scattering did show a much more realistic scatter. But the glow version was still highly effective at getting a quasi SSS effect.
You should be able to utilize the Fresnel effect, I believe that's the one in there, to simulate the effect depending upon the angle of incoming light. But in animations, you're looking at a lot of render time. For a still image, just adjust the light to the effect you want and run with it. Using EP's method in the glow channel would be a great way to use the effect in an animation. If more realistic scattering is needed, simply turn the glow channel effect lower and blend in a bit of real SSS with no map - letting the glow channel effect enhance what SSS is spitting out.
Translucency channel alone, without SSS but perhaps some thick refraction, may also make a worthy substitute to experiment with. This way you have maps usage at your disposal.
That in mind, you could create a conforming innards figure and give a good juicy muscular tissue mapping with glossiness, reflectiveness, and it's own translucency map - which can either allow passage of light or apply the brakes. Then have the main figure use a very subtle transparency with translucency, and the more silky, absorbent properties of or skin. A near invisible fuzz or dynamic fur would even further increase your chances for picture-perfection. It really depends on your equipment, imagination and resources. In the end, I think Carrara can pull it off.
After watching Cripeman's short demonstration of using Global illumination, my eyes are much further open to using that feature as well! He uses sort of the same method as your awesome light map tutorial, but using simple gradients instead of a map. You can then hide that background with an actual scene, if you like - or backdrop.
SSS shader has 3 main parameters:
diffuse reflection
translucency
refraction index
and 2 strength parameters:
intensity
fresnel effect
Translucency accepts any 3-color shader
Looks like a trip to the manual might answer some questions...
...and yeah, I just put the texturemap in the glow channel, reducing it's brightness to about 16%..., then multiply that with the fake fresnel falloff shader from SHADER OPS... (yes, I really do use falloff in every shader, LOL)
but I'm not going for uncanny valley... Just a little more "warmth" to pale skin
one of the pics spreads the falloff wider.... the other also uses falloff but it is more exaggerated and multiplies out most of the glow... I could see using this effect for making figures look tired, older or stressed.
The glow smooths out details and makes the figure look younger and sexier... I think that was my goal anyway... so I am probably not going to pursue real SSS.
It's subtle...
For warmth, I like to use a target object in the appropriate spot on the figure and attach a nice, soft spotlight coming in from underneath. Set the angle really wide, like at least 60, then the angular falloff to around 75-90. If it's 15 feet away, I'll make the distance like 45 and put a distance falloff of around 67 - 90, depending on the light color settings. Attach as many lights as I want, then have the target follow the hip, but not in rotation. I also set each light to "only" shine on that figure, or parts of it. Sometimes I want some shadow, but only from the body - not the clothes. Another light from the appropriate angle will add cloth shadows to the figure.
In CG, we get to fake anything we want. Mimicking true reality might be an easy route, but it's only cheating yourself from all of the wonderful possibilities.
Your figures obviously don't wear a lot of clothes... LOL! ;-P
I look at it this way: In 3D it's all trickery. Even using SSS and GI. They're all just someone's (or a group's) interpretation of what light does and how it reacts. I just prefer to use faster rendering trickery!
So you don't study particle physics first and write a dissertation on the behaviour of photons passing through skin? Here I thought I had the next two decades of my life planned out...
So you don't study particle physics first and write a dissertation on the behaviour of photons passing through skin? Here I thought I had the next two decades of my life planned out...
I thought I had as well, then I discovered nose picking and gave it all up! Now instead of hurting my brain with thinking, I concentrate on seeing how many knuckles I can get in before my finger hits the back of my sinuses. ;-)
Seriously though, I like what Holly posted some time ago (paraphrasing here): Carrara is an artists' tool, not a real world simulator.
Edited to add: Owwww! I think it's time to trim my nails. :long:
Touche! Very well put indeed!
LOL! The renderer debate is a bit trainspotter-ish... I like to say all things are equal and non-competitive, but there are real world problems to allowing a renderer to do most of your thinking for you... That recent biased vs unbiased thread kinda opened my eyes to that.
Actually I think if everyone could render a full scene (regardless of method) in under 2min their results would improve immensely. There are only so many tweaks a night you can get in if it takes a few hours to see the results of each decision.... Improvement will be slow...
The reward of "cheating" is renders that take 1/20th as long...? Yeah, I'm gonna have to keep cheating. :red:
LOL!!! Ouch!
You speak the truth... well, when it comes to Dart and Rosie, anyways. I like it like that, that's all :)Yeah, very true. You can endeavor to get a pretty decent realism sim, if you have the patience, RAM, render room full of refrigerated cores, etc., But why, when we have such excellent options at our disposal? :)_
if your fingers are long enough, you can use that method to manually change your mind:)
if your fingers are long enough, you can use that method to manually change your mind:)
I tried to change my friend's mind that way, but I had forgotten the old saying:
You can pick your friends, you can pick your nose, but you can't pick your friend's nose!
I tried to change my friend's mind that way, but I had forgotten the old saying:
You can pick your friends, you can pick your nose, but you can't pick your friend's nose!
http://instantrimshot.com/
:lol: Stan!
Great points, Holly! That was a fun thread, and that pdf I plugged in really does a great job of explaining what unbiased actually means, insofar as it has nothing to do with how realistic a final result appears, but how the result was reached is what matters. But it is exactly in the way of render time that has me convinced unbiased will soon be the standard. Cycles as well as some of the other unbiased renderers with GPU can perform real time rendering, so changes that are made are immediately visible in a finalized looking way. Such a workflow provides real world light and allows super fast editing all at once, leading to much better final results. With all the details of light behavior worked out for us within the engine, we can focus on other things a bit more than we would have otherwise such as theme, mood, and other details often slighted for focus on other things.
Speed is paramount. Unbiased renderers will all work in similar ways, making unbiased rendering a sort of universal platform. That makes it easy to write optimizations and that may be why we can see such great speed increases with gpu rendering. But biased rendering is different, as there is no way to be certain which types of shortcuts are being taken, it makes sense that we dont have real time rendering of biased engines...its rather difficult to account for those missing values. Sometimes it is easier to do things the right way instead of trickery.
I am a trickery guy, I believe in hand made solutions. but if this unbiased gpu thing keeps going as it is I will probably go for it. A render that takes 1 hour instead of 10 great lowers my electric bill, lowers fossil fuel consumption, and is overall a "greener" experience. There are lots of benefits to speeding up the entire process of image creation. Completed renders in two minutes is exactly what might happen someday. Hoping!
You may be right in the future Rashad, but it seems to me that at the moment it's not quite ready for primetime. I don't want to get into a biased/unbiased debate. One, it's not what this thread is about, and two, because to me it's what works for you. Personally, I have yet to allow a render in Carrara go ten hours unless it's an animation. Yet ten hour renders (or much longer) seem par for the course for Reality (which I believe is an unbiased renderer- could be wrong).
There are other ways to skin a cat, and full GI and subsurface scattering are only a couple ways to achieve a certain look. As has been stated in this thread, there are ways to "cheat," the SSS effect, although my personal opinion is that it's all a cheat, in that it's someones' or a groups' mathematical interpretation as to how light moves through a surface. There is no practical way to measure SSS through every possible surface and write a profile for it. Besides, Carrara, Bryce, DAZ Studio, Poser, C4D, Maya, etc. etc. aren't meant to be real world simulators. They are, each and every one, an art program. Which boils down to, "eye of the beholder."
Again, it's what works for you, and a cheat is only a cheat if you're caught!
I have been waiting for these super hardware enabled realtime solutions.... for like a decade... I remember an Invidia hardware demo at a Mac world... must be 11 years ago? I haven't gotten the blender or the hardware to test caustics... Maybe on my Dell. Ugh... I need about 3 clones and we can just trade notes at the end of the day....
Reality is a nice utility to get DAZ Studio scenes into LuxRender which is unbiased.
But you don't need an unbiased GPU solution to get instant screen update response. There's a new product for Maya, 3DSMax and Rhino that gives very fast realtime updates. https://caustic.com/index.php
They also have what they call OpenRL with an SDK which maybe Fenric could take a look at to see if there's a way of making it work in Carrara.
That looks FASCINATING! Particularly, because Bryce is as far as I can tell a brute force raytracer, that's all it does. Bryce stands to gain a lot from a technology like this. WOW! This would probably be brilliant in Carrara as well, and if combined with the current render engine's particular talents I suspect Carrara will be very far along in catching up to the high end apps.
So there it is. The speed potentially is out there. Just wonder how long it will take for us to get some of that?
Reality seems like a fantastic tool. I hope someday soon it can be made to produce results as quickly as Octane currently outputs. Carrara with an Octane Plugin seems like a brilliant idea and one I thought I'd heard was underway?
That looks pretty darn cool (still watching their video). LightWave has the VPR in their Layout which so far seems to be very similar. I'd love to see something like this make its way into Carrara.
Tugpsx was hoping to work on one for Carrara after he released I believe the free DAZ Studio plugin, but I haven't heard anymore about it. or the free DAZ Studio plugin. I think someone else is working with Otoy on the pay for plugin that has more features that works in Studio. I'm still not sure about what can be done as I'm not a programmer and Fenric pointed out some issues with getting Carrara to work with outside render engines. You can however export OBJ files for use in Octane right now as that has been and is the default for Octane. It works well and the only problems you might run into involve number and size of textures and how big the video memory on your Nvidia card is and whether it's older with the texture number limit or one of the newer 600 series cards that can handle many more textures.
I don't know how Carrara maps the SSS map to the mesh but the basic misunderstanding here I believe is that a map is just wrapped onto a surface of the mesh and effects the mesh directly. A proper handling of a SSS shader would take the density of the underlying mesh into account in relation to the light calculations so that a mesh where two opposing surfaces are closer together would get handled differently then where the mesh faces are further apart. In that respect, adding a solid grey for an area like skin would work as an across-the-board measurement for skin if the skin had uniform scattering. The real difference in the grey values would be where skin has different scattering from the base level, so yes.. there would be no need to use gradients or dark areas for things like hands... the dark areas would be more for areas that disrupt the natural scattering, like bone close to the surface perhaps, scars, etc...