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
Does blender always underexpose images like this? When I opened it in Photoshop, the range was 0-185. To my eyes, that made the skin look to shiny. When I corrected it to use the full luminosity range and a LUT that works well for Daz rendered skin (Filmstock 50—I'm not sure where I got it), the skin looked really good in the face. And that hair is amazing. The body has a problem I've found in my own renders—when you process the render to pull out details, the bumps take on a slightly plasticy look. I think it has to do with the tricky interplay between the dual lobe shader's specularity—getting enough roughness in one lobe and glossiness in the other to emulate the broad highlights of skin with a bit of natural oil on top. I'm curious if Jeff, by producing small, noised up renders is hiding this problem, or if his maps solve it.
This is a great example of where the model isn't realistic enough to pass as photoreal. The skin needs small wrinkles to sell it.
suggested way to say it: "Hi Masterstroke. You could maybe add more small wrinkles to your character, to get closer to photorealism."
Thank you, I will work on it.
I was using Nuke. ACES seems to make everything low contrast. It tries to make images as neutral as possible. This actually makes it the perfect base to post process from.
Here I added some simple contrast. BAM.
As for the body, I think I just set the bump a little too high, lol.
thanks for your input.
Looks so much better now :-)
Hi,
If I understand the question, with dforce clothes it is quite easy to let the cloth drop off rather than stick to a body. Some characters also have nipples morphs and dforce works quite well also in that case. The pictures are a quick sample of what I mean.
Also, you can check the Smoothing Interaction in the Parameters TAB by selecting the cloth and changing this value. I found it helps also with non dforce clothes.
Finally, you can check products, like Fit Control that helps to shape the clothes to do not stick on the body.
I personally think that he said it right the first time.
Certain models always have the 'Daz' look to them. For example, when looking independently at certain parts like the nose, it looks realistic, but other parts are giveaways.
It's about the realism of the model's facial style...but yes, also wrinkles.
More wrinkles, at least wrinkles at all - I get that, but the rest ....what ? I don't understand.
Are you saying, she looks to much... ahm..."DAZ Style"?
In fact, she does not at all. She is unlike all, you can get as Daz characters. Oh, unless you are cherry picking and only mean her nose.
Well, as I said. I don't get your point.
... and yes wrinkles.
Huh..? Are you asking what kinds of things look less realistic?
Well, specifically the eyes and forehead seem to be what's hurting the 'human realism' factor.
But I am no expert.
I personally don't mess with entirely realistic characters, this is as real as I go (see attached).
Huh? what's "hurting" with the eyes and the forehead?
Are you sure, that you have a point here, or are you just being aggressive?
Infact, I think my picture looks really good in terms of photo realism, although not as good as I'd like to have it. I know, still ways to go.
I just don't know, how to achive more in DAZ-Iray. So better to have advises from experts instead.
What bothers me myself most is the hair line here, that looks a bit like a Barbie doll's hair line. But hair is a problem anyway. There is no DAZ or Poser hair yet, that I find really convincing.
Until they come out from transmapped hair I guess this will not change. Really, transmapped hair don't make any sense in pbr rendering. You need fibermesh hair with good shaders.
I marked up your image with places I'd expect to see some fine wrinkles on a woman of that age and weight. As a shape, that's a realistic-looking character, but the parts that you probaby can't control unless you're quite skilled in Zbrush or a similar sculpting program are giving it away. I've had pretty good luck using HD morphs on my customized characters, but the farther you get from the original shape, the more likely they won't have the right details. With this model, which has more body fat than the typical Daz figure, yet not as much as the heavier ones, you might find it difficult to get a concinxing photoreal render without doing some 3D sculpting (and subsequent normal/displacement map exports, since as non-PAs can't import HD morphs).
One thing I think you could improve without Zbrush is the lighting. Overall, it's very nice, but it's blown out on the forehead and chest/colarbone region. Have you used canvases before? If not, I highly recommend them. Canvases are available in the Advanced Render tab and will allow you to export your render in an untoned 32-bit/channel EXR format. From there, you can adjust the lighting in post with more control. If you use the Light Group canvases, one for each light, you'll have even more control in post (plus, you can do bloom in post without the rendertime cost)
For what it's worth I think the pic you did is quite good. The only part I can see where you may want to work is the corners of the mouth... that part looks a little plastic/doll-like but otherwise I have no critiques. Btw, 90 min render??? What are your machine specs? I'd expect to be able to render this in 2-3 minutes.
So-oh,
Did check her again and again.
let's start with
mouth corners:
Sorry there is nothing to do about it. Every single change I add, are altering her character to much.
Actually, her mouth corners look believable- and not doll like to me- and exactley as they suppose to be. So,I won't change that.
To add wrinkles are possible,
but as said, I am not a PA so I don't have HD tools.
I've tried normal maps, but the results are again not quite visible.
Eyes:
Some lower lid Hd morphs indeed look better, but those eye corner wrinkle HD morphs just look misplaced. It needed some, but more subtile and at a better position.
Some laughing wrinkles araound the eyes would be nice. Again don't have the right ones and can't make them.
About canvases and lightning:
I am at the beginning of it.
Some results look very subtile to me. Yes a difference, but not much of a difference. Nothing, that looks more like a "WOW, more photoreal" to me.
Photoshop sucks om HDR images, gimp is better.
Isn't it so, that all the benefits of an *.exr image gets lost, when you try to export it into a *.jpg picture?
Lights:
I used the secondcircle free HDRI lights here with a dome rotation of 25° and increased the contrast a bit and set saturation to zero - all inside DS.
Render times:
My PC is three years old now and I am running (only) one GTX 1070 right now. My character has the Chameleon Rose Peech Fuzz attached, that might slow down render times a lot.
Unfortunately, that's what can suck about this hobby. I have a character I use for a lot of renders that I can never get to look as photoreal as the character below. I don't have the skill to zBrush her and give her that extra bit, nor—as a non-PA—the capacity to bring in any details I made except through render-time crushign displacement maps. I think your character is stuck at quite good, but not fully convincing not because of your failings, lack of skill, etc., but because the body shape requires proeducts that don't exist to take that final step. That said, put in the right environment, with the right lighting, and rendered smaller, she might pass. Jeff's brilliant work succeeds because he limits a lot of factors. My most successful photoreal character works best in studio lighting mimicing fashion shots (there are some examples of that on page 10 of this thread). I havne't tried Jeff's approach, but she might not work as well with that kind of image. Certainly, I have a LOT of failtures.
Nope. At least in Photoshop, you can do a huge amount while in 32-bit mode before you export it to a jpeg. The attached file shows the same render (only using the beauty pass—with .exr files, you can have a different pass for each light, giving you huge control in post) toned differently. In the left one, I intentionally left it a bit over exposed. In the middle and right versions, I kept everything properly exposes. For the right version, I used curves (in 32-bit mode) to get the skin tones closer to the overexposed version without any blow-outs. I cconverted to black and white using a color lookup table (LUT) simulating Ilford 100 film.
as you said, "...Actually, her mouth corners look believable- and not doll like to me- and exactley as they suppose to be. So,I won't change that". That's the point, they look believable 'to you' and that's all that matters so, as you said, don't change it. I was simply trying to point out the one part I felt was 'off' if you're trying to pass it a real photo. As for render times... I'd expect a 1070 to be quicker still... you sure you have it selected in the Advanced Render settings as the Photoreal device?
about render times: Yes it is at photoreal. I think, it is the peach fuzz. The render looks good after 4 minutes at 80%. To smooth out fireflys on that peach fuzz on her face takes so long.
Wow, i see a lot of beautiful pictures and this thread has inspired me so much. I am working on a Asian character myself and trying to make her as realistic as possible. Well as realistic as I can make her.
I am still working on the face textures and then body textures. I mostly painted her face textures. This is what I have got so far:
My two cents for what it's worth; the "peach fuzz" does not add to the realism. It does not look like peach fuzz, but more like male stubble so it looks completely out of place.
I disagree with the poster that told you to add more wrinkles. I think your render looked better without the wrinkles around the eyes. Young people don't have lots of wrinkles. Most of them don't have any wrinkles on their face and when they do the wrinkles are tiny which will be very difficult to simulate with geometry.
I think your original render looks great (except for the stubble ;)), but the brows do not look realistic to me. Viewed from the side at least some eyebrow hair should be visible. Consider adding a fibermesh brow and set opacity to 50 %.
The other thing that I find a bit odd in your render is the necklace. It looks like a low poly wire instead of a rope. Maybe convert it to subd (if it's not already). If it's supposed to be a rope, it's not obeying the laws of physics. The top ring behind her neck is just hanging in the air.
Something that I find in many of my characters (especially ones' that I created by mixing different characters together) is that the eyes can move slightly too far back in the sockets. Your character seems to have that issue. Try moving the eyes 1mm in the Z+ direction.
I've tried lowering the reflections. I think, the firefly filter is at its limits here with the peach fuzz. I have to try some other vendor's peach fuzz, then.
Wrinkles: I'm not happy with those wrinkles, yes, but I don't have the right ones. You are right. Wrinkles can not make the difference between believeble photo real or not.
I'd like some "laughs a lot" wrinkles, not old wrinkles.
Infact those are fiber mesh brows. I corrected some reflections. It is hard to find fiber mesh brows, that don't look like big bars in her face and are morphable to match her face.
Yup, you've got me here. Darn, I've noticed it, but have not been given much attention to it. ;-)
That is an on going issue. for some reasons it keep appearing on and on. It is not something, that I did not adressed yet, but it's like a curse. It allways comes back.
Since those latest posts, I've started to look of lot's of photos with girls at my character's designated age of earley 20s and I figured, that it is not those details, that make it look photoreal - and I mean photoreal, not real world, or human eye real. It is most likely that kind of dynamic range the render and the shaders being used.
Just to test it: It is so hard, to make a render look like a photo. Is it easier to make a photo look like a render, e.g. by cutting down its dynamic range?
No matter, how "bad" a photo is, it never looked like a render to me.
So I am back, where this thread started.
What is the secret? What is that magic potion?
looks nice but not very asian
I think that is because she is probably of different ethnicities. I used a photo reference which labelled her Asian.
Well she looks Japanese to me, have a wide range of looks.
Masterstroke, it looks good. The wrinkles are not a deficit in your art, they are just what others are claiming would slant it closer to photoreal.
I myself don't think it's missing 'wrinkles' so much as facial lines. I think that's what they are actually saying, those crease-lines in the face and chin that we have. I guess you could morph those in, but not even so much, just a touch would sell it. As for the peach fuzz, if it was a render layer I would say lower it to 25% of what you have, because it pops too high in brightness. If the fuzz was softer it would look more real to me.
But, that is just an idea. It's not like you must do it or it is wrong. You have every right to social distance and self quarantine your art from otherpeople's ideas!
I wish I knew the answer to that question. If you ever find it, please do not forget to post it here ;)
It's really hard to replicate all of the thousands of little details that are visible in a photograph.
One of the things that our eyes pick up on is the absence of hair. I think you are on the right track with the peach fuzz. Our bodies are covered with vellus hair. Those hairs are not easy to see on "normal" photographs except in the outline of the face and body. If you look at a rendered image, the outline of a person's face and body is usually very smooth. This is not realistic and our eyes pick up on that immediately.
Part of the problem is that most "skins" only use bump and normal maps and not displacement maps, but they also do not have visible vellus hair. In the picture below I added vellus hair using Strand Based Hair to try and break up the smooth outline of her face. It is most noticeable on her right cheek. I do think it adds to the realism, but I have tinkered with the image for so many hours that I have probably lost all objectivity.
The character is supposed to be in her early twenties.
The hair is by Out of Touch whom I think makes some of the most realistic hair, but it still does not quite fool the eye.
Link to full size picture : https://www.daz3d.com/galleryimage/image/961301/jane---portrait_full.jpg
Whew! Just read through this entire thread. Really enjoyed seeing everyone's renders, and I learned a ton from the conversations. I've been learning Daz for around 2 years or so. Since I sell custom character designs plus stock art, my need isn't so much for photorealism, but rather, renders that look good in book covers and ad graphics. That said, I'm always wanting to push my limits and create beautiful renders for their own sake. And I have wondered how far along some of my renders are getting on the spectrum toward photoreal. If anyone has a free moment, would you let me know what this one gets right and where it misses the mark? I should probably note that I don't love high bump maps due to the crusty orange-peel-like skin it renders, so I tend to gravitate toward HD skins that don't have high bump to begin with. I'm only just now trying to explore and adjust skin settings manually. As a comparison, I attached a second image that shows the same image postworked with my usual process in Photoshop (camera raw, dodging/burning, etc.).
Thanks again for this discussion! Oh! And as an addendum question, there was an earlier discussion on the topic of canvases. I'd love to learn more about that topic. Does anyone know of a specific YouTube tutorial or channel or even a forum post that discusses that in detail, by chance?
Nice work! If you are seeking realism beyond what you have now then you'll need to come up with an environment and then see how the skin reacts in a full environment. the skin should rflect the colors of the environment a little bit and help you to visually decide how much specualr and the best roughness settings.
Masterstroke! Wow man!
You've received some great feedback. Learning how to translate community feedback into useful notations is a skill all its own. I try not to get too caught up in the specific details of feedback, as viewers are trying to help but sadly unless they have the exact same scene file in front of them as you have in front of you.... they can't be absolutely certain their ideas will really improve the render..that is until those changes are applied and the results can be evaluated...and life is too short to try out every single improvement idea you may encounter from feedback. No worries, the feedback can still be extremely useful if for nothing other than providing insight into what different people identify as important.
So here's the psychological part of the study here, which matters both as artists and viewers. It's important to know which parts of a render are working from a realism standpoint, and which ones are not. As viewers we can sometimes take for granted the artistic and technical choices that were effective, and like always we tend only to comment on those things that stand out to us as unlikely or unrealistic, however the things that are right are just as important as the things that may need improvement. I will try to mention some of these under-appreciated strengths in my post.
Learning as viewers to try to look at renders in the way that non-CG/3D trained viewers will observe them is a tricky exercise I participate in regularly. The question is; if I saw this render in a catalogue of sorts with no previous indication that it was not real...would I notice it as not real? And even if I would notice it as not real, would I notice it immediately or would it take me a minute or two of careful study to arrive at the observation that it isnt real? So assuming I was flipping through a bunch of images and this one popped out I think I see the following: Looks REAL!!!
*Hair = Highly realistic....This includes the eyebrows. Perhaps I'll go into more detail later but suffice it to say you are no longer overexposing the hair compared to the other surfaces in the scene. Or to put it another way, you're no longer hitting the highest key on the keyboard for effect sake, you've left yourself some room for more variation nuance and realism in the shading of the hair. It seems you've truly begun to allow the render engine and the lighting within the scene to decide how bright the hair should seem, rather than trying to force a preconceived ideal, and I think the results are great. In fact generally, what I like about what you are doing recently is you are learning to allow the render engine (Iray) to do much more of the work for you rather than trying to accomplish everything with prebaked texture details and arbitrary assumptions about settings regarding how the finished product should look. Seems you're getting out of the render engine's way a bit more is what I'm saying and the results are extremely promising. Please continue in that regard. Let the rendering engine work for you rather than being an obstacle.
*Eyes = Highly realistic (probably the most important thing to get right and the hardest but here you've actually done it extremely well) Super Bravo!!!!!!
*Skin = Highly realistic and I will go further to evaluate the reasons for this
A. Specular chocies are excellent. You seem to understand that skin isnt always shiny in a sharp specular manner unless the lights are really bright and really small. A portait studio type of soft lighting like yours I think should result in a broad and soft sort of specular which you've achieved here nicely!
B. The Normals/Bump or whatever height information you have used for the skin is working well for me too. Subtle, but still visible. This again lends itself to the sense of softness to the skin and realism
C. The expression and pose are also both quite plausible and help create the mood of the scene.
*The vellus hair would look better if it covered the entire body, not just the face. However its also important to note that vellus hairs arent usually perfectly straight, they are curved. This matters because when the strands appear straight as they do in this render it gives the impression of thickness to the strands, a trait most often associated with pubic hair which is straighter because its nearly double the thickness of strands on the head...giving rise to the impression of stubble... the vellus hair should be the opposite, extremely thin and curved....without ever becoming curly as such. Generally I say BRAVO for allowing things like vellus hair on her face knowing a few people might view it as masculinizing, still these "imperfections" if you can call them that are powerful and smart artistic and technical choices in my view.
Also with the vellus and eyebrown hairs I like the high key specular you have used on them. Perhaps it is a little too shiny, maybe a little added roughness could help tone it down a smidge to match the skin, but really, I think its already within a plausible range. The ame type of oils are on the hair as is on the skin, and in microscopic view neither hair nor skin are perfectly smooth...so there ae limits to how shiny hairs should be.
* Some viewers felt there was a need for more wrinkles, both fine and grand, but to my mind this is a slippery slope and almost always where I can spot issues as a viewer. One of the biggest questions in CG regarding human figure rendering is when to bake a wrinkle into the albedo/diffuse color map, or when to represent said wrinkles as mere height map details, or if one needs to combine both baking into the texture as well as some sort of height representation. And this doesnt even consider true displacement. You can have a perfectly uniform surface that when displacement is applied to an SSS type surface, you already get shadows within the crevices and the reddish color shift within the crevices from the resulting SSS on the now non-smooth surface... so again do such details need to be baked into the diffuse image if they are going to eventually be derived from the displacement and the lighting within the scene? I don't know the answer. These types of questions are what I mean by saying that as artists we need to be deciding what to allow the render engine to do vs what we are going to force onto the engine. Wrinkles are really hard...and for me almost not worth it. Personally I don't think those are the issues preventing the scene from looking generally quite realistic.
For my tastes to go to the next step I'd look at modeling my own necklace, as I do agree that the lack of contact with the skin breaks the known laws of gravity and physics and is thus is distracting from a realism standpoint. However from a storytelling standpoint I think the necklace serves to remind us that she isnt nude and that this isnt a sexually provocative type of render...at least not entirely, there's more depth to the ideal here.
Masterstroke: The truth of the matter is...you are using models you did not model, and textures you did not create on said models. You are trying your best to use content created by others for purposes that might not match the original intent of said content. You dont have ABSOLUTE control over the ingredients of this soup, so the outcome is still hit or miss. We are all in this same boat. The more I get into this human skin rendering ideal the more obvious it is to me that true success requires doing as much of the work yourself as possible....take Jeff for instance...pretty much builds everything from scratch to meet his unique and specific purposes. There are likely no skins on the market that do what Jeff needs directly out of the box. I think we should forget the notion that there are intrinsically more and less realistic skin textures, because generally speaking all of the textures sold here are very good, but none are so perfect that they look like real in every single render regardless of anything else.
A few years ago baking every pore and wrinkle into the diffuse/albedo maps was the way to go, because back then with lack of SSS and proper displacement effects you could not count on the rendering engine to do very much of that kind of work for you. But now after all these years, with the advent of Iray and Octane and LuxRender and Cycles and all that stuff, the new lesson is to find ways to let the rendering engine do as much of the hard work as possible, and often that means less detailed color maps and much much much more detailed height maps and displacements. Could you add some blemishes to the diffuse/albedo? Yes. Would they improve the realism? Likely yes. Is it easier said than done...hell yes. So there you go!
Great job please keep it up!
So much great stuff here. Before I go any further I need to make certain you are aware that my feedback is different than what most community members offer, so please feel free to use the feedback that works for you and discard what doesnt. I'm very critical and I often try to consider the thought processes behind the choices we make artistically, so that we can either tend toward or tend away from ways of thinking that might be getting in our way in terms of absolute realism. Realism is hard because there's not much room for artistic commentary. It's either believeable or it isnt. Believeability has to matter more than aesthetic beauty. Okay, that said:
The eyes look GREAT!!!!!! Seriously beautiful! Major points there! However due to other issues the image overall doesnt strike me as photoreal. More in a moment.
The image has a clean look to it, due to the high degree of actualy cleaniliness but also in large part due to the shininess of the many surfaces both foreground and background. Its a very beautiful render! However, to my eye the result is actually more stylized than photo-real due to these specular giveaways. Photoreal isnt always beautiful there must be some ugliness to it...entropy-ish breakdowns...lack of perfect smoothness on surfaces that have existed for a period of time. Your image lacks any and all ugliness...or evidence that time has passed. I'm not sure how you reached these assumptions about how shiny the surfaces should be. I feel as though artistically you've made some arbitrary decisions about reflection related elements that might be working against your goal of photorealism. A few observations.
Generally, ALL of the surfaces in your scene have far too much specular shine to appear realistic. ITs nice to see that you've taken the exaggerated specular ideal and applied it uniformly, as this does create a sense of cohesion between the foreground character and the background elements. Its what makes the image appear to me as stylized rather than realistic. Still, if realsim is the goal, you'll need to lose a LOT of the specular from most surfaces. I'll explain.
Shininess on human skin isnt what we think of it as. Skin even with no oils has a degree of shininess due to its natural level of smoothness. Then there is the contribution of the surface oils. Shininess on skin is derived from anisotropy of light intensities from differing directions. so no matter how bright the lights are, if they are all uniform in all directions then the skin will appear flat and matte, even if there is a very high shininess setting. Indirect light from the sky is a good example of a nearly directionless light. Surfaces that are in shadow from the sun and are being illuminated primarily by the wide angle of the sky, such as her face, should be less shiny than you've made them appear here. The parts of the skin facing the sun are too shiny as well, but due to the angle of view its harder to notice. It's my opinion that your artistic eye has incorrectly assumed the correct technical value for the base shininess of the skin. The problem is in two parts. first, your shininess settings are just plain too high, but also, your roughness settings are also not high enough. Lips are too shiny, unless there is a nearby bright light source I'm missing. If indeed there is a "helper" light of some sort in front of the camera, you should widen it so it doesnt create such tight specular looking reflections.
The top of the building in the background, also too shiny. The overall environment is too clean looking, and also I'm not in love with the way the DOH is operating. Is this derived entirely from Iray or did you paint some of the blurring in post?
The hair is tough. I'm not quite sure how to suggest improving it. Its looks a bit like its perhaps too heavy in shading, doesnt appear light...aha! Got it. With the super bright reflection on the shoulder it gives the impression that the sunlight is very bright, bright enough that the hair should be glowing quite a bit more from the bright sunlight coming in behind it. Theres currently a subtle hint, but based on the other dynamics in the scene it doesnt seem to be enough. Also the hair probably needs even more fly-aways similar to what Masterstroke used in his most recent black and white image. However, you are limited to the way the current hair model was made and the textures that were provided...another instance of how using premade content produced with different goals than yours can tend to be problematic.
Here's something you probably havent considered: Lack of an atmospheric medium also makes everything look too, well, immediate and somewhat staged. Even the overly strong shiny aspects of the surroundings and skin settings would be muted by a nice atmospheric medium to gray things out and lower the contrast, even if it was subtle. You might even be able to add a filter to soften the image and create the appearance of some atmosphere without having to re-render. Right now it is clear that the woman is living in a virtual vacuum.
Postwork:
Contrast seems a bit punchy, as well as saturation. I believe that generally in terms of realism less color will look more realistic, and less contrast as well...in my humble and nearly useless opinion. Fixing the problems at the root of the symptoms could edge this image up to photoreal in no time!
Amazing job, David!
I agree with other posters that the bump settings are too high. You can even see along the collar bone that the bump is different for the chest than it is on the neck. The overexpression of the bump also makes the skin surface appear "harder" than you likely intend, working against the SSS which usually softens the look of skin.
Daz models only use 4096 texture maps for the torso and limbs....which is sadly far loo low. in such a case, you are better off not drawing any more attention to these low res details than necessary., and indeed some amount of bump is necessary, just learn to observe when it crosses the line from being a realism additive to be a realism subtractive element. I think in this case it is possibly a mistake to trust that the baked in details of the height mapping are sufficient. As stated due to resolution limitations of 4096 images, its often wise to add a second type of bump noise that isnt based on a UV map..perhaps some sort of cubic solution, that can allow you to give the skin micro-details that are uniform and that far exceed the maximu detail allowed on a texture map. These are details that will wotrk in tandem withyou roughness settings to allow you to arrive at a final look to the skin that is plausible. In my own work I used both the provided maps as well as either a leather grain or a pores noise type procedural texture. Together they provide me a skin that looks 100% more detailed in close-ups than a skin using only the provided height maps. You need to either use a procedural based on the perlin or voronoi noise types, or you can use a standard noise image of your choice so long as it is truly seamless and is non-repeating enough not to draw attention to itself,
Color Saturation on the skin is far too much. I think the previous one, that you felt was underexposed was to my eye much closer to the amount of color human beings of paler skin complexions tend to look. Probably some middle ground between the two color pallettes would be best.
Realism often less is indeed more.
Hair is outstanding, too bad there's so much Antialiasing pixelation. I think this same AA issue is affecting the skin and the bump settings also distracting the realism. Perhaps the render wasnt completely converged before it was posted to the forums?
Generally I'll repeat some of the things I stated to Masterstroke, in that you need to continue to make clever decisions about what you allow the engine to do vs what choices you want to make for the engine. Its my expereince that the rendering engine is often smarter than I am, and often comes up with results that don't match my expectations at first,...but then after more careful consideration I realize that the egine is trying to teach me something, and i learn to re-assess my expectations to be better in line with actual realism.
Your next challenge is to find some sort of an environment.