Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Nice, a couple things:
Recommend adding some pimples, etc, to the face. I think you'll find this greatly increases realism.
See if you can make the teeth a little more transparent and not all the same color...
Iris color is cool, but if you're going for realism you need to turn down the saturation on those a lot...
Otherwise looks great!
Jeff
!
Yeah its probably a bit OTT. Its the fstop on the camera is was above 22, which means I probably have my camera in the wrong location. I need to start using more accurate camera setups rather than just winging it
Latest... tried to use less 'grainy old photo' style as is my standard and went for a clear, bright scene with two people. Still in draft, but i'm generally happy with it. Thoughts welcomed. _Jeff
Excellent work as usual, Jeff - and a new character!
Did you only use D-force on the clothes, or something like Marvelous Designer / VWD? They look very natural.
Yummy! This conversation is so yummy. Sorry...I don't have many friends.
This observation you've made is both a Yes and a NO for me as far as the parts I can agree with fully and the parts that I may see differently. Basically I conclude that YES, darker skin is easier to fake than lighter skin when it comes to CG rendering. But why? There are two answers.. One more practical, the other more circumstancial and perhaps controversial.
First:The main reason for the relative ease in creating realism with darker skin tones is defined as the emphasis of effects observable to a viewer based on the dynamics of contrast and the limits of monitors vs the limits of the human eye. Yep, basic contrast. Its because of an effect I like to call "Black Magic." Black Magic is real, I swear! The thing with darker skin tones is that we observe the height mapping, bump and specular effects much more readily on darker tones than we do on lighter colored skins that tend to somewhat compete with the specular highlights for "brightest pixel" on the page status. A truly detailed height map with proper roughness and specular settings will always look fantastic when rendered. Under such observationsal conditions, we tend to be less concerend with the types of color variations we might expect from more transparent skin tones that reveal much more of whats happening underneath the skin. Our monitors max out at 256rgb so no matter WHAT you are doing, you lose contrast toward the whiter end of the viewing spectrum as you lighten the scene generally....which sucks because you need that range for your specular highlights, which are sadly often lost on lighter skin tones. This is one area where "camera real" vs "eye real" really makes a significant difference. Dynamic range.
Leap with me here....Your dark skinned image is basically a Black and white render...blasphemy aside......color considerations have been largely ommitted by the high degree of light absoprtion of the darker skin tone in this example, leaving the viewer to focus on the well documented height mapping information which is basic white lines on an almost black background.... and as we've discussed many times in this thread specifically monochrome renders are generally easier to pulll off.
Second: Due to trends in advertising and media emphasis, most of us simply aren't as accustomed to seeing darker skin tones being photographed with fully white lights...so the bar for realism for dark skins is unfortunately..a bit lower than for the ligher skinned counterparts. Alos our real world expericne with a minority grou0 means that many of us simply don;t have the sdharpest imapression of darker tones in outr memory, so we remain somewhat impressionable from this standpoint. Our brains are like deep learning AI...they get better and better at solving for consistency versus inconsistency based on experience. But with a lack of experince, comes and open-mindedness and perhaps vulnerability that can be exploited. That's why its easier to render things realistically that we have the least amount of practical experience with.
For example I could show you a render of a strange thing and tell you that it is a real image of an alien from the planet tekArnaso orbiting a binary star system in the Andromeda Galaxy... and chances are if it was physically consistent...it would be difficult to disprove without a control sample (a real photo) to compare it to.
Just as you lose specular highlights on lighter skins, you will lose It is actually for the reason of discerning height maopping and believebale specualr that I often encourage artists wirking on human skin to work with varying skin tones, not just stick to a singular area of skin tones. What you are seeing here, is how much more beleiveable we have gotten with height mapping than we have mnay other as;ect of character rendering in Daz3d. So yes, you are fully correct...we can somewhat cheat by using darker skinned folks than lighter skinned ones, and again this is only because we can allow height mapping _bump, normals, displacement) in cobination with specualr mapping to make some really cool looking humans. However with lighter skin tones the specular opften competexs with the diffuse/albedo, and specualr appears to get lost and skin overall seems less beliveable.
It should be noted here, that you specular settings on this lovely dark skinned model are very much too high, unless she literally just took a bath in oils. Still this is an swesome work!
I'd like just to add a note here that's important to understand and I see not everyone seems to get it. That is, developing a photoreal image and developing a photoreal 3d asset are two different things. When you use "tricks" such as a specific material that only looks photoreal with a specific light set, that is not a photoreal 3d asset.
I feel the need to stress this because I see some tutorials around intended to be "professional" that are teaching to adapt skins to lights and/or lights to skins to get them photoreal. A photoreal 3d asset works fine in any (realistic) light condition. That is, photoreal materials do not depend on lights.
Also I see some tutorials teaching to use the uber shader outside of its recommended limits, always to reach for more "realism". The uber shader is very flexible to allow for conversion of 3delight materials. So it can also mimic specular materials that are not pbr. Beware that when you go outside the uber shader recommended limits you go outside pbr materials, so you enter the "fantastic" materials domain that will never behave as realistic in different light conditions.
Being in the trenches of materials conversion myself. That's all. Not intended as futile criticism, but as a reminder or warning for anyone sensible to the matter who wants to develop true photoreal 3d assets.
Hi Padone, great insight. Do you know how to convert the Disney PBR specular (from principled shader) into DAZ? I've been trying to get the fresnel effect in DAZ to look similar to blender but can't.
@davidtrune The uber shader is more flexible than the principled shader so I wouldn't expect issues converting from principled to uber. If you may post an example of what you're after may be I can help.
Just basically converting the specular part of it. So you know how there are just two values, the specular (IOR) and roughness? In Uber there is IOR and roughness, but there's also a "weight". I don't know what to set this in relation to the roughness. Would it be like 1 - the weight? Because I know the roughness has an inverse relation to the fresnel weight.
@davidtriune If you mean the uber glossy layered weight, that is intended to convert 3delight materials and should be left alone for pbr materials. So you shouldn't use it to convert from the principled shader. As for conversion equations that may apply I'm not at my workstation right now. I'll give it a look as I get home and let you know what I find out since I'm interested myself.
http://docs.daz3d.com/doku.php/public/software/dazstudio/4/referenceguide/interface/panes/surfaces/shaders/iray_uber_shader/shader_general_concepts/start
I was talking about the Top Coat weight not Glossy Layered Weight.
I also read it's similar to this node setup in blender, if you've ever seen it before.
@davidtriune The uber top coat weight is certainly to be greater than zero to produce any effect. This relates to the clearcoat channel in the principled shader so it has nothing to do with the inner layer material. Then the uber top coat is extremely flexible and to find out what are the settings to match the principled clearcoat may take some time.
Indeed in the diffeomorphic plugin I didn't have time yet to find the exact relation.
https://docs.blender.org/manual/en/latest/render/shader_nodes/shader/principled.html
Wouldnt being a "top layer" not matter if it's the only gloss you use? or what is the difference between "inner layer" and "outer layer", besides one being rendered before another?
@davidtriune The top coat is intended to mimic an extra layer of refractive material to add to the underlying material. From a physical point of view it is certainly not the same thing to use it instead of the material reflection. Then it may happen than in some contitions you may get similar results but this is another matter.
An alternative is to use dual lobe but that isn't supported by the principled shader. That's why the diffeomorphic plugin uses the standard bsdf nodes for dual lobe.
https://bitbucket.org/Diffeomorphic/import-daz/issues/25/
https://bitbucket.org/Diffeomorphic/import-daz-archive/issues/321/
Straight DS Iray render. No postwork exept resizing..
Diffuse skin maps were added to the emission shader channel at default values. Character is in a box prop. Camera headlamp is up to 4.
Some of Secondcircle's exposure values added to the scene. No canvases used here.
Yup, I know, some hair and lashes issues stink here ;-)
That's true but only under the assumption that the used render engine and shaders are 100% correct in the first place, which is not necessarily the case. I certainly don't feel that is the case for skin rendering in Iray. So then the only option is to try and push something further by trusting your eyes and winging it.
Hello everybody,
Finally, I feel ready to join this discussion! I've been following ever since I started fiddling around with DAZ and was pretty inspired by the advice exchanged here :)
I would like to post one of my artworks for comments and criticism. I applied an HDRI and one emissive surface to simulate sunlight only. I also spent a pretty long time on manually editing the model's skin to add asymmetry and nunanced tones.
I am quite satisfied with it - but then again I browse this thread and see so many awesome, awesome renders of characters that just...'pop' a lot more, for a lack of better ways to say it. More crisp details on the skin, textures looking more realistic.
I feel like there's something very foundational and basic about DAZ that I just don't fully get yet and would love some advice!
(Wasn't sure about this boards rules on nudity - hence the black bar ;) )
@bluejaunte The uber shader is not excellent for human skins from a technical point of view, since a dedicated sss solution is not supported and it is replaced with a generic volumetric solution. This is why it is difficult to get a good human skin because you don't have easy dedicated parameters for it. Nevertheless it is certainly true that pushing the uber shader outside the pbr specifications will only get you non realistic materials, that may very well look good under certain light conditions, but that's not as a true photoreal 3d asset is supposed to work.
https://www.daz3d.com/forums/discussion/comment/5425546/#Comment_5425546
Nice! maybe use fibermesh eyebrows and the sclera is a bit white to my eye... but good job.
There's no such thing as a true photoreal 3d asset. CG is always an approximation of reality. Otherwise you'd have to model everything under the skin, every vessel, vein, fat pads, muscles and bones because that is what the renderer would need for true realism. Not a bunch of crappy make-belief SSS maps.
Take random walk SSS as an example. This is a fairly recent development and you can read how it's "leaps and bounds" over previous methods. So in Iray we don't have this, does that mean we should stick to the supposedly one correct way to achieve good SSS even though we know this is apparently not the best way to begin with? Or should we go outside the specs and try our best to get it to look better? Of course we'll probably fail in this case, but the point stands.
There is no 100% photoreal renderer out there. Otherwise everyone would use that one, or if all renderers were 100% photoreal then they'd all look exactly the same. Look at how complex VRay is, why is that when all you have to do is follow the one given formula and hit render to get the one truely correct result? It's a whole bunch of fake hocus pocus, that's why.
This is quite good, but you're having a little problem with the hand on the sheet. Did you use the denoiser? Her details look a bit smudged to me.
Infact, the eyebrows are fibermesh brows.;-) Don't know, where those super white eyeballs come from. They look very realistic in other renders. I had to darken her teeth a lot, in order not to have them look like tictacs. Maybe that extra bright camera head light is the reason.
I'm afraid black bars don't work https://www.daz3d.com/forums/discussion/3279/acceptable-ways-of-handling-nudity#latest
Okay! Took your suggestions and worked on the eyes, skin, and teeth a bit more (i.e. hours of fiddling).
I must say, I'm proud of this render. This is the most realistic one I've created yet. Even so, I'm always looking for tips and feedback!
3 things I'd note:
1) There's a harsh transition from the iris to sclera heres a real world example (also showing some nifty translucence)
But the key is that transition, in general anytime you see a sharp line its going to look cg.
2) The nasolabial fold is very soft for such an expression, even on younger people when we smile things get a bit creasy (generally in my expirience the crease will be strongest towards the corner of the mough and then can soften as you approach the nose) also the area below the crease (between it and the upper lips) follows and tightens against the gums, where in yours it seems to have pulled forwards. I tend to try to make the expresion in real life and then move back and forth between it and neutral to feel where all the musles pull (with my hand dont do this in front of people you will look real silly) Expressions are hard though.
3) The gums look off. Similar point 1 theres a crisp transition and theyre a very uniform in color generally gums get slightly lighter closer to the teeth as the flesh is thinner and much like teeth theyre pretty translucent. this one is a massive nitpick mind you, you're not going much gum in most photos and even when you do, very few people are going to be inspecting it. I think the only real reason I even noticed it is because the teath actually look quite good indeed. and theyre night next to it.
Great critique, thank you!
How would you recommend smoothing the transition? I suppose simply blending the two textures together?
are her skin veins on diffuse or translucency maps (or other)?
Translucency, which is at 0.90, I believe.
Hello,
Oh I'm sorry. the image is deleted already I see. My apologies, won't happen again.
I cropped it down to a headshot now. My problem is that I *think* light works, and that I *think* I gave the scene a realistic tone but I am just completely lost when it comes to that smudged look, as @Leonides02 noted. I would love to get rid of this and get a clearer image that shows more skin details.
I did work with DoF and an F/Stop of roughly 9, a Focal Length of 85 but did NOT use a denoiser of sorts. Again, no headlights, just HDRI and an emissive plane. I'm truly at a loss as to what I'm missing for a crisper, richer skin.
(And yes, I realised that some parts of her hair are a bit off but that can be easily fixed with a bit of postwork ;)