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.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, I wasn't saying that there was anything wrong with them or that it wasn't nice skin. Only that you can't get results like yours with point-cloud occlusion, which is supposed to give more accurate representation of subsurface than standard 3Delight, according to their own documentation, without doing a lot more tinkering with either the Ubersurface settings or the point-cloud occlusion render settings, or both. That's it; that's ALL I was saying. I hope you didn't think I was saying anything about the quality of your settings or images; I really didn't mean to come across that way. Just that point-cloud works very differently than you'd expect.
Here's what I mean. On the left is David 5, in the FleshForge Render Room from ShareCG. Uses Wancow's light setup at full strength. D5's default texture uses Ubersurface -- not HSS and not US2, just plain Ubersurface, and his settings are whatever the defaults were.
Second image I turned down UE2 to 50%, because it was clearly too strong for the Render Room, and the reflections were blowing David out.
Third image is point-cloud occlusion, Wancow's lights with UE2 at 50%. All point-cloud settings were left at their defaults as well.
Fourth image is point-cloud with UE2 at 10%, because it was still too strong. At which point I gave up, because it's clear that the entire light setup is too strong for point-cloud in that setting.
He's also got a noticeable yellow shift, but I have absolutely no idea where that's coming from. The render room is reflecting white; his specular color is that weird muddy green that's his default (in both channels, even), his subsurface is pink, and translucency is off. Wancow's lights are either pale blue or white, and the occlusion light is white. So wherefore yellow?
wancow and evilded777... would any (or both ^^) of you mind posting screenshots for beginners in this spheres (and non native english speakers who sometimes don't really get all the words), like me to better understand? It sounds very interesting and I think I could learn a lot for my renders.
Thank you for considering :)
Screenshots of what BlazeMystEra?
VWrangler, Nope, not what I was thinking. I didn't take offense. I bet you CAN, actually.
My settings are designed to look pretty much the same, regardless of the lights.
Why would you use UE2 AND point cloud occlusion? They are doing the same JOB, and the point cloud, holy crap, is sooooo much faster. All I changed here is to swap out the UE2 for the scripted renderer, tweaked the settings on the render job, and away we go. Is still a little brighter than my original image, but damn that rendered in 1/4 of the time or less.
can you provide a link to this Flesh Forge, my search fu is weak.
Render settings in the scripted 3Delight are the same as regular render settings in Studio.
Point light down to 40% and a light blue, same as UE2 previously.
Cloud shading rate down to .2, clamp on (I dunno what this is for, but I am assuming it is a feature relative to trace distance in UE2)
Hey wancow, Just FYI:
The link to this thread on ShareCG just sends you back to the page it is on there at ShareCG. You have to manually C&P the link to get here.
whup, sorry... my thoughts jumped over the important part again. Screenshots of the surface/shadertab of the skin :)
If you left the point cloud settings at default then I think you got the point cloud simple occlusion light on top of all the others and in this case I wouldn't be surprised about this result. If that is the case you can disable the simple occlusion light within the point cloud rendering settings, UE2 alone still benefits from a point-cloud setup without that light. Or, like evilded777 said, you can kick the UE2 light out and just use the simple occlusion light, depends on what look you want to achieve.
A few scattered ideas - a more coherent version is going to come in the long treatise I'm writing. With lotsa pics to prove my points. But you gotta wait some 'cuz real life, y'know.
No piccies in this post 'cuz my internet connection hates me tonight. But I haz links. Very recommended ones. Read on, and ye shall find.
-- CRUCIAL: SSS scale 0.1, SSS shading rate 1 - ask theSea if you don't trust an unknown like me, it was him who guided me towards this in our old lab thread; also consult 3Delight docs re:scale conversion! Up your scale, and the shader thinks you're feeding a doll to it instead of a lifesize human! ---
--- CRUCIAL, pt. II: if you're using SSS at all, all your lights must cast shadows, IBL must use AO. Otherwise you'll get light rays where they don't belong (c) theSea in the old lab thread, again --
1. The diffuse maps we're using have SSS already "built in" - for one simple reason: we can't take SSS out when we photograph a real person (ask Mec4D if you don't trust me). Only the blood-based portion of it if we photograph a corpse...
2. This is why, if we're using blood-based colours for SSS, it's better to drop the diffuse strength somewhat (or to desaturate your maps and stick the diffuse in the SSS colour, but that won't really work for the default UberSurface because its SSS colour map control is broken).
A cyan tint in the diffuse colour multiplier is a neat addition if you want a paler-yet-scattering character.
Read this, BTW: http://www.bossanovatech.com/Articles/Article P&G on skin translucency - 2006.pdf - might spark interesting ideas. Gotta love solid science!
3. The original UberSurface is a frellin' beast to work with. Its SSS controls are limited, and the colour parameter is extra sensitive at the physically correct scale. Pale, very pale yet not too pale - try these: 249-255-221 or 234-251-255. Only allow 255 in the red channel, if you want a Martian (or a living human out of a zombie/alien diffuse map).
And yeah, go wild with the SSS colour if you do want a cool alien.
4. Unlike its cute li'l sister UberSurface2, US prefers energy conservation to be respected re:strengths of diffuse and SSS channels, otherwise it does tend to produce overexposed images. Diffuse strength 75%, SSS strength 25% - works more or less OK.
5. Ditch the "SSS maps" unless you're obsessed with making ears glow. If you are, paint your ears white and the rest of the map a shade of grey. Other than that... look at how deep light actually scatters in skin and then do tell me you need those mythical maps (bones getting in the way of scatter?? unless you're holding a candle in your palm, it means someone failed their anatomy class) http://orion.bme.columbia.edu/~hillman/Skin_Imaging.html
// sorry for causticity, but this is one of the worst pet peeves of mine //
Skin is skin is skin. Physically it's made of the same cells all over your body, and they scatter the same. Your diffuse maps are going to handle the melanin irregularities like moles, palms/soles etc.
5. But you NEED tons of intricate specular work (and preferably blurred reflection as well, 'cause Fresnel DOES NOT attenuate spec in the original US, only the reflection) to make it look prettier. DAZ3D's inhouse materials have always excelled in the specular department - even at the times of the default "DAZ Material".
I really hate using the default UberSurface for skin, because I'm no specmapping wizard. I prefer subtle-yet-important simplicity that working fresnel attenuation of specular provides. //and yeah, I suggest a different workflow for US2 that does not formally respect energy conservation parameter-wise, but that's a whole 'nother discussion//
Here's a couple of my relatively recent renders that use US2 and oldschool maps: not that they "show a lot of skin" (I don't do that!), but, like, a litmus test. If you go "whoah, that's horrible", then just block my posts and forget everything I ever wrote.
- a more realistic one ('cause the textures are realistic) http://www.daz3d.com/forums/discussion/14967/P105/#253653
- a more stylised doll-like one ('cause, well, The Girl's textures are painted) http://www.daz3d.com/forums/discussion/14967/P150/#259648
I use UE2 set to AO+soft shadows with maps either made in Bryce or converted from Openfootage.net, and a single distant light (raytraced shadows on these 'cause it's a "shader" light from the SDK and it's fast enough for them)
...other points gleamed from the thread that I'd like to comment on:
a) Now, blue specular. It's a GOOD THING.
Skin is dielectric (that's why it scatters and fresnels an'stuff). For dielectrics, real specular is white. BUT:
"In order to get a good non-metallic result (neutral specular color), we need to neutralize the specular reflection using the inverse color of the diffuse... Because, in RGB, whites and grays contain all colors. Since they have all colors, they will also contain the diffuse color, and once you add this portion of the diffuse color, contained in the grays of the specular, on top of the original diffuse, the result will give the material a metallic aspect"
Source and piccies: http://www.manufato.com/?p=902 - a very useful page overall. Really, read it, folks.
b) Wancow, could you explain the physics-based reasoning behind your decision to use SSS on fabrics?? The way I see it, I don't believe fabrics truly scatter in the same way solid substances do - they consist of multiple threads discernible with the naked eye. If anything, light PASSES through the fabrics (US has Translucency for that), and tiny fibers scatter surface diffuse light at specific angles (this is what Velvet is really for, it's not for skin unless it's a baby covered in downy fuzz or something...). There's nothing really going BELOW (=sub) the surface of fabrics.
Now leather would scatter classically, but not fabrics...
c) "Enclosing characters into some primitive"? Only necessary if you're using crazy raytrace distances for your AO or IDL: "crazy" is anything over 25 cm.
Actually I'm using "crazy" not because I'm questioning someone's intelligence, no - I'm not THAT evil. In my parlance, "crazy" means "adding unnecessary overhead to render time".
Yet, this time crazy distances are also simply not realistic. Try playing with actual objects IRL and note the distance at which you start noticing the reflected indirect tint.
Crazy distances are only justified if you're using UE2 in GI BOUNCE mode (when it DOES NOT provide light by itself, only bounces your other lights against surfaces). If you are using UE as an actual source of light, there's the map doing most of the fake bounce for you.
Now if you're using raytraced reflections, you definitely need an all-enclosing environment, but a simple primitive's going to be boring in reflections... at least, stick something large and colourful behind your camera.
d) UE2 and point-cloud occlusion. If you actually read the docs for 4.5 (those lists of "new features" or something, that's where it was), you'd see that UE2 is now designed to TAKE ADVANTAGE of point-cloud. Drop it in and set it to AO with soft shadows. If you set it to "ambient only", it will wash out a lot of your point-based occlusion, BTW. And setting up a thousand oldschool lights to simulate envlighting... good grief, that's crazy. As in, unnecessary wasting of render resources.
Lower the occlusion distance in the scripted renderer settings, as well.
Alright folks, I have a big headache, so I'm really really sorry if I sound offensive or something. It's not intentional. Just try thinking in a foreign language when your brain won't cooperate with the simplest tasks... it's no fun =(((
Wow, mustakettu85, that's some food for thought.
A few things
SSS scale and shading rate: I learned a lot from theSea, but he and I disagree here. From the 3Delight docs, you'd think the SSS scale ought to be .1, but in practice it doesn't work so well. The shading rate, as noted by 3Delight can benefit from higher values on translucent surfaces. I won't go above 48, but I rarely find a need to go down below 24, in practice.
You are right about the diffuse maps having SSS baked in, a lot of times they have specular baked in too... this stuff is hard to overcome (why do all the newer textures have such RED legs?)
All over the shadows and AO. Gotta have it, and you gotta have some sort of global illumination.
That bit about specular colors gives me pause. It contradicts some of my own reasoning, but I am willing to learn more.
Good stuff.
It does work for me. Especially in shaders with better controls, like US2 or the shader mixer SSS brick (which is set to the correct values by default, and it's one of the reasons it renders so well without much hassle at all). I used to doubt this as well, but when I was doing scale tests for that thread, I saw it clearly that 0.1 looks the most natural.
And as the skin is not as translucent actually as that coloured stone the chess figure is made of in the 3Delight docs, the shading rate needs to be dropped.
Surely if you're using the higher conversion factor (scale 1 and above... let's call it "figurine scale"), your geometry is treated as smaller and hence much more translucent by definition, this is why you could up it all the way to 128 (like in those DAZ3D mats for Elite samples that come with 4.5) and never notice - but it's even more of a hassle to get pretty looking skin this way than just finetuning the capricious colour control in US at the 0.1 scale.
When I was using the "figurine" scale, I was continually and painfully struggling with the geometry impacting the way SSS looked, all of the time - this, for instance, is an image I'm pretty happy with - http://mustakettu85.deviantart.com/art/New-beginnings-278508676 - but it was a nightmare with all the trial and error involved. And good grief, did I run into artefacting ("banding") big time - see, figures are relatively far away here! So they need lower shading rates to accurately shade their small area they occupy in the render, aka more operations per pixel - before I realised that the SSS shading rate actually could be dialled down a lot without my computer dying. And that's a scale of 2 or 3, right. Actually the scales are different for The Girl and The Dude! All because of their different geometry. Polygon count and placing becomes truly important at these "miniature" scales...
Here are my US2 tests from that old thread - very basic setup: US2, D3 hi-res maps, SSS preset "skin3", bump min/max 0.02, spec strength 15% with bump maps, diffuse 40%, SSS 60%. The lighting is UE2, AO+soft shadows, a distant light with DSM in front and a spotlight behind the ear =)
Quoting myself from that thread:
"At the default US2 SSS scale of 1 everything looks nice, but the banding is evident on his shoulders and scalp."
http://forumarchive.daz3d.com/postimages/origimage_2_3383506.jpg
"With a scale of 1.5 the banding is more subdued but it's still there if you look hard... //I upped backscatter boost from 2 to 3 here//" - the thread was also about this major obsession with the glowing ears
http://forumarchive.daz3d.com/postimages/origimage_3_3383506.jpg
"Scale 2 makes for no visible banding, to me. But I think I like 1.5 better //the difference is quite subtle to my eyes, though//"
http://forumarchive.daz3d.com/postimages/origimage_5_3383506.jpg
"And a scale of 0.1 produces this insta-alien effect =) Sorta like those fish people in the first Andromeda season."
http://forumarchive.daz3d.com/postimages/origimage_4_3383506.jpg
...and then theSea made my day by telling me to lower the shading rate.
http://forumarchive.daz3d.com/postimages/origimage_1_3383693.jpg
This looks the best, to me. I've stuck to 0.1 ever since and never ran into any issues.
Re: Spec colour - somewhat counterintuitive, true. I used to grumble about the "unnecessary" blue in a lot of mats, too, but then I came across articles like the one linked, did some experiments and realised that a lot of the time, it is indeed very useful.
From same thread: shader mixer SSS magic done all wrong, but still looking good =)
Really, really wrong. In the treatise I'm writing, I'll be describing the network I am using now which is done correctly (re:placement of the SSS brick) - this placement leads to excruciatingly slow render time and generally wrong shading one too often, when the shader is any more complex - but this Magic Mistake is fairly useful as is.
The SSS brick is buggy in DS3 in the sense that you can't save a scene with it O.o - but otherwise it's alright.
http://forumarchive.daz3d.com/postimages/origimage_1_3405762.png
...and the results:
"The lighting in the test renders is two spotlights with DSM (key and fill lights). The textures are Thorne&Sarsa;'s Parabella and the paler diffuse one with eyeliner is LynX (the bump is Parabella's). The morph is Parabella." - no UE
"The skin shader is only applied to Aiko's head surfaces (head and scalp)." (lips use a slightly more specular version of same shader; neck is default "DAZ Material")
http://forumarchive.daz3d.com/postimages/origimage_3_3405762.jpg
SSS should not be at 90% with low diffuse. You lose all the detail...and try backlighting it. :lol:
FleshForge Render Room prop. It's not really meant to be used the way I was using it, but it was there, and I was lazy, so. I tend to use it in lower light situations, and I've used it to do stuff I think of as "chiaroscuro" style, lots of sharp contrast in shadows and light. I use it in preference to posing backgrounds sometimes because it is fully enclosed. That can make camera movement a bit weird sometimes, but it means that it can be easier to control lights and effects.
Wait, I just did... where's the catch? =)
You're absolutely right about the detail, though. A bump map helps immensely, but still... Anyway, with a nice shader, this type of setup has its place, I think. It's not photoreal, but tangibly soft. Looks great on toon-type figures, I think.
Sixus1's HER, no maps, just US2 with white diffuse and "skin4" SSS preset; diffuse strength 10%, SSS strength 90%. A distant light at 100% intensity as a backlight, a spotlight at 75% as key. Raytraced shadows. No UE.
PS SSS scale 0.1, SSS shading rate 2.
PPS Spec strength 150%, pure white, Fresnel 98%, falloff 2.
Nice mustakettu
Bare install DS, Basic Genesis with default skin, some settings tweaked. UE + 3pt light setup.
Render time 3m47s
[Edit] I updated the picture because I missed a setting I thought I had reset (specular set to a bluegrey, changed that) which threw off all of my other settings. If anyone is interested in the settings on this, I have a writeup here.
http://www.daz3d.com/mindvision-g-d-s-
MindVision is using some fantastic lights, and I am unsure but think they are being rendered in DS with 3Delight.
Does anyone know?
If they are DS lights are they in a set sold by MindVision?
Those renders, in my opinion are some of the best I've seen.
I'll often adjust textures by adding colour values either under or instead of the diffuse texture.
Here is a scene using DAZ default content. Nothing genius about it, but it's quite typical of the types of adjustments I would use for 3Delight renders. Normally, I wouldn't really use DAZ default content for a "serious" scene, but I'm using default stuff here so anybody can load it if they want it (well, so long as your version reads .duf files) and have a play around.
Link to dA post for the scene file http://fav.me/d6495qe
Wancow, I think your light values are turned up way too much. Particularly when considering conserveration of energy.
The first is my materials and that setup.
For the second, changed the UE settings to Occlusion with Soft Shadows, Maximum Trace Distance to 125, colors to 75% (192,192,192). Occlusion strength is 100%, samples 128, color is 0%. The lights were toned down too - colors 75%, strength to 50%.
For the third, I reupped the fill lights strength to 80%.
I generally just use UE2, a distant light and an Uberarea light (for fake indirect lighting).
Shadows could use probably softer, but I didn't change them for these renders.
Some light study shots.
Textures are different, but the material settings are the same (HSS).
I started with Wancow's spotlights, transferring them from DS4 to DS3. The only difference is the y position (I didn't factor in the height of the null parent). I'm only using the main and fill lights (no backlight). I then replaced the two lights with a distant light, positioning and orientating it until the shadows position match. For the last shot, I used an UberAreaLight (works only in DS3 I think), to get the same amount of 'fill'.
I generally like softer shadows and find the spotlight shadows to be unrealistically sharp.Even with 100% softness, it's hard to get softer shadows with spotlights. In contrast, even with 0 - 5 % softness, shadows with directional lights are not as sharp as spotlight shadows. I find UberAreaLight the best for fill/indirect lights, since not only can you control the light (color, strength, falloff) but also the shadows (intensity/color). Notice the darkest area on the earlobe are not as dark (I've set the shadow intensity to about 75% for the area light). It would've been hard to do so with just the base lights and or materials/texture tweaking.
Don't know how omAreaLight compares, but I think you need more than just the basic lights.
Did I read it right? One hundred and twenty five centimeters? One surface influencing the light affecting its neighbours at over a meter away... I mean, really... How lifelike is that?
If you're running AO, it makes your image one too dark. And you have to up the UE quality setting way too high (the more area AO covers, the more samples etc it needs to be noise-free).
If you're running IDL, it overlights your image - for that same reason, you can't really reflect diffuse light to one meter away...
Once again: you only need max raytrace distance of several meters if you're using UE to do the GI bounce for you (when it provides no light by itself but bounces your other light sources).
The IDL mode of UE does the colour bleeding, the diffuse tint you get from one object onto another. \
These are different things.
Have you tried using DSMs on your spotlights? They're softer than raytraced shadows by default. For a spotlight DSM to be noise-free, you'd need about 16 shadow samples (either in the general render settings, or in the light's own settings if you're using "shader" lights or UberSpot-type ones).
Never comfortable with DSM shadows. Primarily, because the shadows never seem to be where I want them to be.
Edit. Added some renders to illustrate why I chose 125 for Max Trace Distance.
First render is with the default value (500). Which is way too dark. Second render is 125. Third render is 50. In the 50 shot, you can see a dark area on the ground around the ball. For me, 125 looks like the best balance. it may not be the most accurate, but it works good enough for me.
No, not using maps. If you were following my posts you might have noticed the particular guidelines I was following was bare install, nothing added but the prop, so no map, thus the extra lights to get the results. I agree adding a map would reduce the number of lights and have other benefits, but doing a proper map adds it's own level of complexity that I didn't wish to incorporate at this level.
I have a bit of a different take on things then Mustakettu. It's a bit long winded and I don't want to bore anyone who isn't interested, but if one does care to see it, they can follow the link I previously posted. I will say that I tend to use UberEnvironment to set up a 'base' level of ambient, maybe a touch more. I then get a lot of the more specific lighting from individual lights. The particular formula I use is a bit involved and varies from image to image somewhat.
@wowie, very nice example. Your mention of UberSurface is very applicable here. I probably would get better results using them then the traditional 3pt light setup I used. I am just now getting comfortable with UberSurface lights myself and really like the results I've gotten so far.
Skin is not dielectric, just like water is not a conductor. In electronics, water is used as an insulator actually, and dielectrics are specific physical entities with specific properties. To say skin exhibits dielectric properties when combined with oils and other substances is somewhat correct just as saying impurities in water, which is a fluid medium allows for conductivity is correct. The reason I get nit picky about this (which I try to avoid) is because incorrect conclusions are being drawn from incorrect statements.
[Edit] I stand corrected (did some research,) Skin is dielectric. It occurred to me I had heard that in biology class long ago. So my apologies. My statements about light and the way it does not fit into neat rules as much as one would likes still stands, per my discussion on the other thread.
Just to clarify btw, there are three categories, not the two as mentioned: conductors, insulators, and dielectrics. On a base level, there are many more true dielectrics then insulators iirc, but what constitutes a dielectric is a point... in electronics, they are generally referring to an item that specifically refers to a level of dielectric properties which function at a basic minimum level of dielectric characteristics. It is easy to mix the general category of dielectrics which make up the world around us with specifics that only really effect things when they reach a certain minimum level of dielectric characteristics, or that operate on a graduated scale of some arc configuration, rarely in an on/off manner.
The lights are Adam's pure white set that he uses for material testing, only change I made to them is to set UE2's shading rate to match the same setting in my render settings.
The texture is the same V4 one that wancow used in the second post, the only changes I did to what was loaded from the PZ2 was to change the spec settings on the skin, gloss to 65%, color to 128's and strength to 15%, I then dragged the skin into ShaderMixer so I could get access to the controls of the skin lighting model, and changed the bright red scatter color to a more natural skin tone.
It looks good, but I notice a band of light at the back of his head.
I like these as a starting point, I'm going to play with them
I noticed that. But if you have UE on, it still lights everything, even if you stick no map into it. Even if you turn AO on. It still floods your scene with light.
It's just that the map - any map - makes the UE shader much more predictable.
I guess you are a super lucky person all in all, if you never run into artefacts and other displays of misbehaving technology even when you're breaking the rules big time.
So... you not using AO??
But UE without a map and without AO does the exact same thing as simply turning the ambient channel up on all the surfaces. Evenly.
The only use I personally ever found for that overwhelming flat ambient is toon rendering.
Isn't it a bit too much of an extra work, flattening out the whole scene first and then trying to fix it with multiple lights?
If you want repeatability, you could just use one of the sample maps that come with UE, one of those that everyone has. There´s that very neutral grey-n-white one, I think it's called Softbox.
Wait, but then why do cell phones go wanky after being submerged?? Is it not because of water (which is, as we´re taught here, unless purified, always containing some electrolytes) forming unwanted electric pathways? Or are you talking of purified water, in fact?
Just to clarify btw, there are three categories, not the two as mentioned: conductors, insulators, and dielectrics.
So you don´t have semiconductors in the English-speaking countries?? Here in Russia, it's (as I said on your thread) conductors and dielectrics aka isolators, and then there are semiconductors closely related to the latter.
Well here, we tend to use this definition for dielectrics, and no I didn't further derail the conversation by bringing in semiconductors as it didn't apply. If you notice, dielectrics, at least in english speaking countries, refer to a specific subset of insulators, that exhibit particular properties.... I think that is what I've been trying to say all along, though perhaps not very well. I apologize, I haven't actually done electronics for a while so I'm rusty. And yes, pure water is an insulator at the base elemental level, that was part of my point. Actually, any water is, it's the impurities that move within the water... hmmm... forget it ;p
Off to make an apple pie while listening to the sound of one hand clapping.