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
For what it's worth, having a dedicated processing pipeline for simulating subsurface scattering effects in human skin is itself a hack away from reality since - in reality - human skin is subject to exactly the same laws of optical physics as any other substance that allows light to pass through it. From a purely techincal standpoint, Daz Studio's Iray uber shader (with its triple-layer volumetric scattering effects) is a much more faithful-to-reality way of modeling SSS skin effects than popular rival methods such as depth mapping or texture space diffusion.
OOT hair os great when it comes to shader and geometry, but doesn't look like the daily average girl next door hair style. Looks more like a hairdresser's promo, party style or catwalk.
AprilSYH hair is pretty good as well.
I didn't have time to do any good renders that are deserving of this thread, maybe later this week.
@Masterstroke, if you use a photoreal hair, the image might be very convincing! Looking very good!
Since i started to use daz3d i tried to achieve a photo-realistic render effect, in the past i first tried Reality plugin + Luxrender then when i bought a new computer with a good graphic card i switched to Iray engine.
I am not sure if i am qualified to post on this thread but this is one of my last render it would be nice to receive some suggestion to improve the visual quality.
You perhaps might like to adjust the dimensions of your image attachment so that it doesn't distort when posted inline.
You're right my fault! I made the post quickly and i haven't resized the picture :-)
This is a lovely render! But I'd say the photorealism falls short because of how "ideal" your Diana Prince looks. In fact, it's quite "comic-book like," which isn't a failing at all.
If you want to improve the photorealism I'd say focus on giving her a little asymmetry and reducing the gloss on her lips.
The hair looks quite good, by the way!
In pbr everything is a "dedicated pipeline". We have nodes for volumes reflections diffusion scattering etc. Each one dedicated to model a specific aspect of real materials. Thus a node dedicated to sss is really no exception. Anyway that was to explain why it is diffcult in iray to get good skins, replying to @bluejaunte.
So we seem a bit confused here as daz PAs. The uber shader is the best for skins but it's not correct anyway. So we have no choice but tweaking around. That of course I disagree.
As for the uber shader way to do skins it just uses a translucency layer with true volumetrics behind it. There's really nothing special you can do the very same in any engine with translucency and volume nodes. But again iray misses a dedicated sss solution that usually gives better and faster results that's why most production engines implement it.
edit. For completeness I'd add that I may have to correct myself since skin shaders for iray do exist. But I don't know if they are just uber shader presets or if they are real dedicated shaders.
Thanks a lot for your suggestion, i'll try to add some imperfections to her next time, probably i tried to make a movie edited character and we know how much the make up hide the natural defects.
Anyway i post here some older works..
Curious - does the shell fit just on the tips or the whole hand. Could it morph with a pose or a long acryllic nail?
This is false. Unbiased PBR rendering engines (like Iray) actively favor generalized rendering solutions for achieving specific visual phenomena (like SSS in human skin) because that is how light physics works in reality.
Light doesn't interact with human skin the way it does because it is human skin. Light interacts with human skin the way it does because of the composition and structure of the organic materials that make up human skin - just like with any other physical object. Having dedicated rendering pipelines for achieving specific visual phenomena in specific substances/material surfaces is the exact opposite of PBR (Physically Based Rendering.)
You are describing biased rendering engines here - not unbiased rendering engines like Iray. They are completely different animals from a design (and implementation) perspective.
Unbiased = real is marketing hype. Functionarlly, everything we do is faked when not dealing with very simple surfaces. Real world items have tremendous complexity that can't be captured in a few maps, and that's especialy true with something as complex as skin. If a dedicated skin shader is more effective at creating skin that looks real to our eyes across more lighting situations than a general shader, it's a better realistic skin shader than one that pretends to be more physically accurate but doesn't achieve the same results.
@RayDAnt Generalized vs dedicated solutions has nothing to do with unbiased vs biased. As well as using nodes rather than a general purpose shader. Indeed cycles is an unbiased engine the same as arnold and octane and they all support a dedicated sss solution and nodes.
Then I feel it is important to understand that we're not speaking about iray itself, but about the uber shader that's a different matter. The uber shader allows for non-pbr materials to convert 3delight shaders. For example specular colors are non-pbr nevertheless allowed. Then iray also has shader bricks that allow for non-pbr materials as well. So you see that having an unbiased engine is not enough to always get pbr materials, nor generalized solutions make things any better. It rather depends on the way shaders use the engine features.
A different hair this time.
Looks good!
I took the liberty of running it through photoshop to apply the ACES color space. It takes away the slight red tint / glow to the skin. What do you think?
This is great. A layperson probably will take this for a real person. It is very, very close to photoreal. I think the eyes, skin and proper exposure interaction all work to great benefit here.
I did a lot of shader tweaking here, she looks good in almost all lighting conditions :)
why is she so reflective
Render engine biasing has nothing to do with marketing hype. Biased/unbiased is specialized terminology for whether or not a rendering engine uses recursive algorithms like raytracing. Rendering engines like Iray are said to be unbiased because they are capable of simulating a potentially infinite number of individual light ray samples for the creation of a single image of ever increasing internal quality.
But it has a lot to do with PBRs - and everything to do with unbiased PBR's in particular. Because the objective of those is to simulate reality in an as faithful fashion as possible.
As per each engines' documentation, Cycles, Arnold and Octane all currently utilize Random Walk as their premier "dedicated" SSS solution. According to the Arnold whitepaper (pages 32:9-10), "random walk" is a volumetrics based method of SSS simulation (based largely on methods proposed in this 2017 paper from Pixar) developed orginially for Arnold as an upgrade from the less effective "Christensen and Burley" volumetric approximation method. Meaning that they are all in the same ballpark (at least in terms of underlying engine functionality) as Iray when it comes to achieving high quality SSS effects.
Node vs uber shader design (see this paper from Arnold for an excellent rundown of the relative pros and cons of each) is a user experience dilemma and has no direct bearing on underlying engine functionality.
You misread what I wrote. I didn't write that Unbiased is marketing hype, I wrote the idea that it, or any "physically based" render is more real than than biased engines is marketing hype. I deleted the rest of what I wrote because it's not very on-topic to worry about biased vs. unbiased: that's probably another thread, if not another forum all together (cgtalk seems like a more appropriate venue for technical comparative disccusions of various rendering approaches).
You are conflating physically based rendering (PBR) with render engine biasing when they are completely separate things. Every rendering engine is classifiable as either a physically-based biased renderer, a physically-based unbiased renderer, a non-physically-based biased renderer, or a non-physically-based unbiased renderer based on its programming. You could certainly quibble with what it is exactly physically "based" really means (I see tv episodes "based" on true stories all the time...) But that has nothing to do with biasing.
I would dial down the gloss, but she looks good.
Is the background a photo plate? You positioned the camera well, if so!
PBR does not equal unbiased, this is a major misunderstanding from what I have been reading over the past several years. Iray is a Biased PBR Engine, meaning that it is physically based, but it is not truly unbiased. Physically based means that energy conservation is observed, and that metals are treated differently than diaeletrics. The issue isnt the "what, but the "how." These PBR considerations can be done while still employing biased methods, such as shortcuts in the way scattering rays, reflection rays, shadow rays and other effects are carried out. Octane, Cycles, LuxRender etc.. these are fully unbiased engines, and follow all of the pbr steps, as well as doing so in a manner that doesnt make any assumptions to speed this rendering along. Mere PBR engines can cheat. Take Keyshot for instance...amazing looking renders, very photorealistic and fast rendering, but not unbiased, because it doesnt need to be. And in the right hands the renders will look visually satisfactory even if they arent perfectly accurate, visually the faults are not observable. The differences are hardly worth mentioning in most contexts, however the unbiased engines are supposedly more reliable and predictable in their output
What the heck is bias anyway? Which is rarely discussed I find. One example of a possible bias is the way feeler rays are fired from the camera into the scene. One could allow a truly random ray firing and eventually they will have rays reaching each pixel in the scene at least once. However since the camera is firing rays in 360 degrees along all three axis, there will be many many rays that are fired away from the scene, behind the camera, toward the sky...etc...places where there are very few likely surfaces for light to be absorbed or reflected from. This method is slow but statistically it should cover all pixels and light influences even those not visible directly in front of the camera... if enough rays are fired...to iron out the noise. This fully random ray firing method is lovely, accurate, but highly inefficient. However one could bias this approach by sending most if not all rays to the areas where they are most likely to hit a pixel such as directly forward rahter than in all 360 directions, and that will save time while producing a result that might not be too far off from the fully randomized ray approach.
Iray is biased, but it is accurate enogh as PBR engine that we can still hold it to the same standards as we would Octane and Vray.
See here for a full technical definition of what unbiased/biased rendering means. Iray is a fully unbiased PBR rendering engine because it uses a combination of Metropolis light transport and bi-directional pathtracing as its core rendering method (see page 5 of this document.)
Thanks for these links. Based on what I am reading it states that Iray is only unbiased if indeed both MLT and Bi-directional are both enabled at the same time, but as you have read they don't both have to be enabled at the same time, and I am curious if users have control over exactly how and when these two modes are utilized. I'd love to know more.
This question came up some years ago and I remember reading this post by the Iray development team at the time:
https://blog.irayrender.com/post/142742319456/is-iray-an-unbiased-renderer-can-it-be-used-to
Anyhow we don't need to get too deep off into this part of the discussion, because it isnt important. the only thing that matters is does iray provide a close enough approximation of SSS to produce results that are competitive with octane and Arnold and what not? To my eye the answer is yes, irays "volumes" are probably good enough to compete, biased or not, so we need not give up the fight to get amazing results from irays skins.
One thing to keep in mind is that this question was asked/answered in 2016. The Iray whitepaper I linked to was published in 2017, and Iray has gone through extensive engine upgrades since then (eg. rendering of caustics has now been fully integrated into the primary rendering pipeline.) Fingers crossed for an Iray whitepaper update some time this decade with fresher information.
Yeah, this is mostly academic. Once you get to a certain level of complexity in PBR unbiased renderer design they all become essentially the same (hence why the imagery they generate tend to look more and more similar as they are improved - everyone's aiming to hit the same bullseye: reality.) Apart from how they expose their capabilities to shader code, that is.
Last night Iray render..
Again not quite. And you still confusing iray with the uber shader. Random walk is a dedicated volumetric solution for sss, it is different from the general volumetric solution used by the uber shader.
A general volumetric solution is of course implemented in those engines as well and it is intended to be used for volumes not for sss. Indeed the uber shader can be implemented with translucency plus volumetric nodes I did it in cycles. That is different from the random walk and it also gets different effects. For example the translucency desaturation in random walk is not possible with the uber shader.
Anyway I feel the need to stress that your references are always top quality so thank you for them your comments are very interesting to me.
Also, my goal here is not to fall in an academic discussion on unbiased or bpr or engine comparisons. It is rather to sensibilize the PAs to use the uber shader at its best for pbr rendering. Since they tend to use the non-pbr features of the uber shader that are only intended for 3delight conversion. As explained in my first posts with @bluejaunte.
The uber shader guidelines for pbr rendering are well reported in the uber shader documentation but I'll get a few of them here for anyone interested.
http://docs.daz3d.com/doku.php/public/software/dazstudio/4/referenceguide/interface/panes/surfaces/shaders/iray_uber_shader/start
1) Avoid the pbr specular glossiness and weighted modes since these are difficult to manage in a pbr way. The pbr metallicity roughness mode is much more standard and simple to use for pbr materials.
2) Always use either 0 or 1 for metallicity. A map for oxidized metals is allowed to have small transition values.
3) Do not use the glossy layered weight since this is only intended for 3delight conversion, and always use a white glossy color for pbr materials.
As for skins it's a more complex matter but I can give a couple of advices based on my experience with material conversion.
4) Translucency alone is responsible for back scattering effects, you don't need volumes for that so a simple skin can have thin walled on.
5) The chromatic mode is buggy do never ever use a 100% green channel.
Is the AMD ProRender Engine physically based & unbiased? Cycles? Eevee?