PDA

View Full Version : Lightwave 10 and 11 skin nodes



isaac3d
01-03-2012, 09:58 AM
Hi,

Lightwave 10 and 11 have a new skin shader node, simply called Skin.
Can anyone explain the main differences between the different skin nodes; Fast Skin, Simple Skin and Skin.
Advice as to appropriate situations to use these nodes would be welcome.
Strangely the number of variables available in the supposedly more advanced Skin node is lower than either Simple or Fast Skin.

I would like to render as realistic skin as possible and I realise that this will involve quite a lot of effort, but if anyone has written a tutorial to give a head start in this area then that would also be very welcome.

I have searched the internet and found only tutorials relating to the Fast Skin and Simple Skin nodes from LW 9.x
It seems a little strange to me that the new Skin node has received so little public attention, unless it's in fact a waste of time. I hope that this is not the case.

I have also read that multiple layers of skin can be used, though I don't see how, unless it is by making multiple Texture Layers and putting a different skin component on each one. That would be possible in earlier LW versions too, of course.

Any advice is welcome.

RebelHill
01-03-2012, 11:51 AM
fast skin is interpolated, so better for stills, not so great for animations. Simple skin is more brute force, and can be used easier with animations. Skin is good for animations, and renders with VPR... thats about it.

Surrealist.
01-03-2012, 12:06 PM
There is some good information given to me here on using Skin:
http://forums.newtek.com/showthread.php?t=124762

Basically from my initial testing there really isn't anything worth bothering with for animation other than Skin. It is fairly straight forward out of the box.

This is a great read where a lot of alternate solutions are presented:

http://forums.newtek.com/showthread.php?t=122729

The Skin documentation is in the 10 docs. I ran across it the other day but now I can't remember where it is. Apparently not where you'd think it was because at first I couldn't find it.

If you get familiar with nodes you'll probably see how you can set some things up with Skin. And if you find it in the docs that will help.

Greenlaw
01-03-2012, 01:16 PM
RH pretty much covered it for differences.

FWIW, the new Skin node is my default choice these days. VPR compatibility means I spend a lot less time dialing in the settings. Skin can also be relatively fast if you optimize your settings.

At work, if we have an insanely tight schedule, we might do our SSS shading as a post effect. The technique involves using an SSS shader with only three lights in the pass, colored red (front scatter), green (back scatter), and blue (spec). By feeding this data into s custom Skin tool in Fusion, we can dial in the look we want. We might also take advantage of the normal channel if we need the spec to look more correct for the enviroment. It's a 'fake' SSS trick that's not necessarily accurate but it renders quickly out of Lightwave and gives us a convincing result in compositing.

G.

RebelHill
01-03-2012, 01:57 PM
Thats an interesting trick GL... might have to do some experiments around that idea.

Oh yeah... should also have mentioned before, re you "realistic" skin comment.

The difference of a "bog standard" render off either SSS, or plain ol lambert surface is actually quite subtle, noticable, ofc... but still subtle. What really makes skin look real is the textures (including things like bump), and the lighting... as well as this, the single thing that makes any surface look real (be it skin, wood, metal) is the specular/reflective attribute.

This means the best workflow approach for getting the most realistic results will be to ignore all materials to start with, and first light your model. Once uve a got a grey thing that looks like a solid, real object, then add textures... when that starts to look real (from a perspective of surface detail), then its time to get stuck into shaders. With the skin materials, since they include a diffuse/reflection model, its probs best to make it coloured black, so as you can make ur specualr look just right.

Then pipe the colours in, and start working on the actual sss itself, depth, visibility, etc.

isaac3d
01-04-2012, 09:38 AM
Hi,
Thanks for your replies and suggestions. I have followed up some of your links and worked on your suggestions and I think I'm beginning to see where to go to get some good results from the Skin node.
Of course, I'm cheating, I imported the V4 character from DAZ3d and I'm also using textures (Dublin by Danae; Renderosity) as inputs for various nodes (see the attached image).
To get good results I had to crank the Skin Diffuse Weight up to 250% or it was too dark, Scattering weight also works best at 100% rather than default 60%, I found. Over 100% Scattering weight doesn't seem to make much difference with the other settings I have. The same is true for Roughness.
What makes a big difference is increasing the Scattering Samples from default 8 to 32 (64 is even better), this gives a much smoother image but slows the render down.
Here the face textures only have been set up and the render (1024x760) took just over a minute.
I wonder if there is something else I can do to improve smoothness (ie reduce graininess) with less time overhead.
Overall, I'm pleased with the image and think this set of variables could produce some nice renders.

RebelHill
01-04-2012, 10:01 AM
Few things...

First, 250% diffuse is WAY too high. Notice how your eyeballs appear dark? Thats cos the skin is overly bright. Now.. ideally, this shader is tuned for working with colour corrected linear workflow (do a forum search for more info)... basically, jsut set ur CS options to the sRGB preset.

Next... ut taking ALPHA from ur b/w spec map... wrong... thats the images alpha channel which (assuming the image doesnt have cutout/transparent areas) is gonna be pure white... use the LUMA output.

You should probs up your scattering distance too.... 4-5mm is good, assuming ofc that your head is modelled to a real world scale.

As for graininess... explore your AA and adaptive sampling in camera properties... You can use AA to clean up a noisy image. I usually use Skin with only 1-2 samples, and let the adaptive aa clean up the mess. Thats a whole world of stuff to be getting into though, so for just getting to grips with using skin, u might find that keeping the samples high as u have is gonna be easier to start with. But the other method can deliver smooth results with less render time once u get the hang of it a bit.

isaac3d
01-04-2012, 12:32 PM
Hi RebelHill,


First, 250% diffuse is WAY too high.
I would tend to agree but it works and I haven't found a way to get round the problem that setting it at 100% gives a skin which is much too dark (see the attached image).
As for the eyes; I do agree but I haven't done any work on them in these images.
I'll look into the linear CS issue.


Next... ut taking ALPHA from ur b/w spec map... wrong...
True but I tried Luma as well and actually thought Alpha looked better :)
The top left image below has the Luma of the Spec map into Specularity... it works too.


You should probs up your scattering distance too.... 4-5mm is good, assuming ofc that your head is modelled to a real world scale.

V4 is a Poser model and in Poser world she is about 2 meters tall. The PoserFusion plugin imported her at about 0.5 meters tall (don't ask me... ask Smith Micro!! ;) )
I tried setting Scattering distance (depth, I assume) to 5mm and got the image shown bottom left below. It is too much, but what is interesting is the strange artifacts on the nose and ears. I noticed that the pattern is the same as the black artifact I get when I leave "Exclude from Vstack" checked, see bottom right of attached image. No idea why that should be so.


As for graininess... explore your AA and adaptive sampling in camera properties... You can use AA to clean up a noisy image. I usually use Skin with only 1-2 samples, and let the adaptive aa clean up the mess.
The second image shows a render with the Scattering Samples knocked back down to the default value of 8 but with Adaptive Sampling checked in Render Globals. This seems to work well and is slightly faster; about 40 seconds:thumbsup:
AA (antialiasing) was always set high, but that doesn't seem to effect graininess anyway.

Thanks again for the help. Further suggestions are always welcome :)

Carm3D
01-04-2012, 12:50 PM
Can you see from the eyes that your subject isn't getting enough light? I agree with RebelHill.. Put your diffuse back to 100% and increase the lighting.

isaac3d
01-04-2012, 01:48 PM
Can you see from the eyes that your subject isn't getting enough light? I agree with RebelHill.. Put your diffuse back to 100% and increase the lighting.
Unfortunately this does not (entirely) solve the problem. I normally use an Image world light with an hdr probe image on 80% and two area lights fairly close by as front lights and one as a back light all set to 50%. I have tried cranking these up to 200% and much higher but it just causes artifacts and washout spots to appear and does not solve the problem.
Perhaps you can see from the neck which has a normal image texture applied that the lighting is adequate. Sure I can improve the lighting but so far anything I've tried with lighting doesn't solve the problem.
I have tried replacing the area lights with different types of lights, but that gives no joy either. I could increase the overall light by adding several more lights, but I'm not sure that won't also cause artifacts and washout, and will surely increase render times.

isaac3d
01-04-2012, 02:20 PM
I have just tried increasing the brightness of the texture image I use as input for the Skin Colour and Scattering Colour and that made quite a difference. I think that with some playing around I may be able to get the image light enough that the Skin Diffuse Weight can be brought down to normal levels. Perhaps it is just a case of reducing the amount of red? (What I want to avoid is just washing it out till all the detail is lost.)
Obviously I can do that in photoshop (which has the advantage of doing it once per texture, but it is harder to fine tune), but I wonder if there is a modification node (or combination of nodes) I can use to increase brighness from a texture or reduce the red component... whatever?

Carm3D
01-04-2012, 02:25 PM
I do not have LW10 yet so I cannot speak to the behavior of Skin.. But if the eyes do not have any special materials on it, then they are not getting enough light.

As to your rendering times issue.. I would refer you to this (http://www.youtube.com/watch?v=oKfD9TGv7Oo) video (and it's 2nd part (http://www.youtube.com/watch?v=O9hoAPWMgFg&feature=related)) I posted on YouTube a ways back.

RebelHill
01-04-2012, 02:28 PM
The reason ur getting blown out areas with strong lights, and dark skin with more regualr lighting, is because this node is designed for linear workflow... to use it properly u do need to use the rgb preset in cs options. (though u can uncheck the convert 8 bit to float, this is of no real help most of the time and just uses up memory).

Also what I meant about the plugging the alpha in, is that the alpha is just 100% (white), so no different to having nothing plugged in and the value just set to 100.

isaac3d
01-05-2012, 01:00 PM
Thanks for the videos Carm... love the fart at the end :thumbsup:

Rebel... will report back on the linear CS when I've tried it out.

Greenlaw
01-05-2012, 01:22 PM
Additionally, check the colorspace setting for the texture in Image Editor. Individual images can be in different colorspaces in Image Editor so it's possible that the setting is incorrect there.

isaac3d
01-05-2012, 01:51 PM
Here are some renders (screen shots of them) with different Colour Space settings.
It is clear that using the correct CS for input and output is important (as Jarrod Davis says in his video :) )
Keeping Skin Diffuse weight at 250%, changing the input/output CS to sRGB gives better results.
Reducing the Skin Diffuse Weight down to 100% is not a dramatic change but is actually better in my opinion and does not make for a much darker image as I had expected. It is slightly darker, but I can easily compensate by increasing the light by a few percent.
Changing to the correct CS setup seems to have solved the problem... thanks guys :thumbsup:

Greenlaw wrote:
Additionally, check the colorspace setting for the texture in Image Editor. Individual images can be in different colorspaces in Image Editor so it's possible that the setting is incorrect there.

I checked and saw that the individual images have CS set as "Default", which I presume means Linear. If I have set input/output to sRGB in preferences do I also need to change individual images to sRGB? Or does "Default" actually mean "What the preferences say" ?

isaac3d
01-05-2012, 01:55 PM
oops... posted to Quick reply...

Here are the images...


Just for clarity as the text I attached to the first two images has gone missing...

First image is rendered with Skin Diffuse Weight at 250% and CS preferences all default Linear
Second image is rendered with Skin Diffuse Weight at 250% but CS input and output set to sRGB
Third image is rendered with Skin Diffuse Weight set to 100% and CS input and ouput set to sRGB.

Greenlaw
01-06-2012, 12:09 AM
I checked and saw that the individual images have CS set as "Default", which I presume means Linear. If I have set input/output to sRGB in preferences do I also need to change individual images to sRGB? Or does "Default" actually mean "What the preferences say" ?
I think Default means it will use whatever is selected in Preferences for that image type. If Float is set to Linear, float images will be Linear. If 8-Bit Files is set to sRGB, then the default for regular images is sRGB. Unless I'm mistaken, whatever you select in Preferences becomes the Default.

FWIW, I prefer to set everything to Linear. For me it's just easier and more predictable to work with.

Waves of light
01-06-2012, 01:15 AM
I use the preset drop down 'sRGB' in the CS tab and that seems to set up everything displaying correctly on my screens and outputs.

RebelHill
01-06-2012, 05:22 AM
nooo, gl... dont just set everything to linear, thats CS off basicaly, lol.

Yes, the sRGB preset, as u have things in 3rd image is what u want. You can see how the skin and eyes are even in their illumination, a bit dark overal perhaps, but that just needs to bump up your lights. Also, being in coour correct space wi mean that you can push your lights, and you wont get the burnout that you'd otherwise get when not using cs workfow.

Waves of light
01-06-2012, 06:55 AM
nooo, gl... dont just set everything to linear, thats CS off basicaly, lol.

Which is what I was doing. Not very intuitive, but I was lucky to get help from Matt in another thread.

Greenlaw
01-06-2012, 09:46 AM
nooo, gl... dont just set everything to linear, thats CS off basicaly, lol.

Yes, which is fine with me.

Greenlaw
01-06-2012, 10:07 AM
I'm going to reveal my ignorance for a moment and clarify: On my own projects, whenever I use Lightwave's CS, the results are always a hassle to use in other applications and I don't have the time to deal with it. I might change my opinion if I ever took the time to learn how to use it properly but for now, with CS disabled, I'm getting predictable results for working in Fusion and other apps, which is okay by me. That's strictly my opinion though and I'm not suggesting that others follow my practices.

FWIW, Lightwave's CS is a hassle to use at work too. We enable sRGB there but unless we save from our custom imagesaver, the gamma always need to be compensated for in Fusion and other apps. Not sure what's up with that but up until we revised our imagesaver, CS sure made us nuts. I can only guess that we're not using CS properly in the Box either. (This is just my opinion but I don't think any of the artists in the Box really understands how Newtek intended CS to be used, and most of these guys are a lot smarter than me.)

G.

geo_n
01-06-2012, 11:09 AM
CS was created in lw 10.1 to degamma/linearize everything. That's the same thing I learned in vray. If you don't degamma all inputs then it is "wrong". Still a matter of opinion though because I've seen great work even without lwf in vray or lw. Working with CS in lw just makes it "right". If you turn it off or default then its the same as lw 9.6.
The only issue I've found with lw 10.1 cs is the AA, AS. I don't know maybe I'm missing why I'm not getting good clean edges in lw 10.1 cs enabled without using higher AA, AS than what I normally use in lw 9.6.

Greenlaw
01-06-2012, 04:44 PM
If you turn it off or default then its the same as lw 9.6.
Yes, which was fine with me. When you're rendering and compositing in float space anyway, I'm still not sure what the big deal is.

G.

geo_n
01-06-2012, 07:06 PM
Yes, which was fine with me. When you're rendering and compositing in float space anyway, I'm still not sure what the big deal is.

G.

Your textures, lights, colorpickers are set to gamma 2.2 while the lw render engine before lw 10.1 interprets them as linear because lw renderer work in linearspace. They need to be degamma'd or converted to linearspace because "wrong" input is taking place even before hitting F9. To degamma in lw 10.1 you set them as srgb. Default or off just lets the "wrong" input to take place.
Working in float in compositing is fine but the textures, lights, colorpickers, etc were already "wrong" in lw in the first place. "Wrong" if you wanted linearworkflow which is one point working in float.
But again I've seen great work without it.
Matt has a simple video explanation about it. Vray has tons of tutorials about it since way back and it probably started linearworkflow in commercial renderers.

RebelHill
01-06-2012, 07:17 PM
put simply... the first 2 of the 3 head renders above show what the basic deal is. Both same diffuse/surface values, but the one with cs diabled is more contrasty. Its not that the image itself is more contrasty, its the response to light of the shaders, darker shadows, and overblown highlights.

Previously, rather than moneky around with de-gamming everyithing, we've worked with weird lighting, diffuse, spec values, etc, things not addding up to 100%, as one way of cheating round it.

But now its built in a lot of shaders, adn lights, and so on, are designed to work with it... Basically all u need to do is set to the sRGB preset (disable 8bit to float, not reli needed), and then work normally. Ur final renders will be saved as sRGB, exactly the same as they would have been with everything disabled, so fusion, afx, etc, should show them no different than they appear in LWs viewer... u may need to set the imported footage in ur other app to srgb colour profile, if its not set to it already.

U can ofc, the continue to use the benefits of lcw in ur compositor if it supports it. Dont know about fusion, but in afx, u set the poject to 16/32bit turn on linearise working space, and then all transfers modes, etc behave in a linear fashion, avoiding the overburn u get with non lcw there too, banding, etc.

jasonwestmas
01-06-2012, 07:27 PM
Make sure your spec and alpha maps aren't being set to srgb cs turn them to linear in the image editor when using sRGB color space. Same goes for Normal and all other maps that aren't intended to show color values.

And yes you will notice that the Skin Material node is half as bright as the simple skin node. I was told by the developer that this is because Skin is more physically accurate or some junk. Soooo you will have to crank it up some to match what is going on in Fiber filter and other shaders but not likely to 300% diffusion.

For realistic indoor lighting use an area or sphercal lighting value of 1milllion% with a inverse ^2 falloff. Then decrease or increase as needed.

Cageman
01-06-2012, 07:59 PM
FWIW, Lightwave's CS is a hassle to use at work too. We enable sRGB there but unless we save from our custom imagesaver, the gamma always need to be compensated for in Fusion and other apps. Not sure what's up with that but up until we revised our imagesaver, CS sure made us nuts.

Nuke allways assume that the data is in linear colorspace, making renders that doesn't have been put through a CS-process, come in with too high gamma. Using CS in LW, where everything set to sRGB except for output RGB and output Alpha, which are set to Linear, Nuke loads them in and display them exactly as how LW F9 renders looks. Oh and... there are some types of textures and images that you will have to make local overrides on in order to work correctly. A normalmap saved as a PNG, or JPEG for example, will be loaded in as sRGB, but it should be Linear. There are several .hdr images that, since they are float files, will be loaded into LW as Linear, but they allready have a 2.2 gamma encoded in them, so, in such cases, needs to be changed to sRGB so that 2.2 gamma is stripped out before it goes into the renderengine.

When we rendered with LW9.6 (where we never actually worked with linearizing the data); in Nuke we had to add nodes to remove gamma, because Nuke assumed that the data was linearized and by default it adds 2.2 displaygamma. Since we had to add nodes to remove gamma, it added some extra complexity when using buffers such as Refraction.

When I use Fusion with data comming from LW where CS is used, I add a Gamut node and set it to display in sRGB. I've noticed a huge difference with lights/hotspots etc when working with CS; lights and their falloffs works much better and makes more sense, and darker areas have more information in them.

I understand that it is different and will yeld different results (converting an old project will require a lot of lighttweaks etc), but if you start a new project and stick to CS, you will notice some pretty good advantages. But yes... as I said, you have to add gamma if you use Fusion.

EDIT: Another note on Fusion; if you set your output saver to sRGB, it will work out nicely in Fusion. But if you have a mix of Nuke and Fusion for compositing, it is better to have the saver set to Linear, and add a Gamut node in Fusion for display purposes.

geo_n
01-06-2012, 08:08 PM
There are several .hdr images that, since they are float files, will be loaded into LW as Linear, but they allready have a 2.2 gamma encoded in them, so, in such cases, needs to be changed to sRGB so that 2.2 gamma is stripped out before it goes into the renderengine.


Really? How do you detect if an hdr has 2.2 gamma encoded in them? I always assumed they are clean and set them to linear.

Cageman
01-06-2012, 08:26 PM
Really? How do you detect if an hdr has 2.2 gamma encoded in them? I always assumed they are clean and set them to linear.

They look washed out in the image display. They also produce a much more flattened look when used as lightsources.

geo_n
01-06-2012, 08:31 PM
Right, thanks for the tip. I always loaded them in imageworld so its hard to notice.

Greenlaw
01-07-2012, 03:39 AM
Make sure your spec and alpha maps aren't being set to srgb cs turn them to linear in the image editor when using sRGB color space. Same goes for Normal and all other maps that aren't intended to show color values.
Good to know. At work I've been doing that with normal maps but I didn't think about doing that with other 'non color' images.

G.

Eric Walters
03-18-2012, 10:10 PM
Isaac
I'm wondering how this worked out-and in general if it is easy to use the new skin shader to make the character look good. I'm going to use the same Danae texture in my render example. The scene is lit with a snow hdri and an infinite light as a back/rim light.
Can Poser 11 or 10 equal this render? If so can someone post images and node setups?


oops... posted to Quick reply...

Here are the images...


Just for clarity as the text I attached to the first two images has gone missing...

First image is rendered with Skin Diffuse Weight at 250% and CS preferences all default Linear
Second image is rendered with Skin Diffuse Weight at 250% but CS input and output set to sRGB
Third image is rendered with Skin Diffuse Weight set to 100% and CS input and ouput set to sRGB.

Waves of light
03-19-2012, 01:15 AM
A normalmap saved as a PNG, or JPEG for example, will be loaded in as sRGB, but it should be Linear

Did not know that one!

Some great tips and examples, many thanks.

Ricky.

Greenlaw
03-19-2012, 01:24 AM
Yes, that's an excellent tip! This issue (with normal maps) really messed me up last spring until some of the texture/modeling guys at work pointed it out to me. I haven't checked to see if this is spelled out in the manual but if it isn't, it really should be.

Quantenschaum
04-09-2013, 07:35 PM
113455
hi guys, i only recently switched from LW 10 to 11.5 and of course i checked the "Skin" node right away. it's a cool node, but in my opinion it's more for doing pixar-style skin or general sss surfaces. i like the roughness feature and i wish that "simple skin" had that too, but in the end "simple skin" still rules where realism is concerned.
my "simple skin" pros: it has 3 color slots giving more texturing possibilities and it actually does render in VPR too. good looking skin is mainly a product of epidermis + subdermis, with only a hint of diffuse color, that's why i find the 2 color slot approach of "Skin" somewhat limiting.

the head in my render is a highres scan which came with a 4k color, normal and a 32-bit displacement map. based on the color image i created an epidermis and subdermis image. with the "nDo normal" photoshop plugin i made a "cavity map" based on the default normal image (for pores and wrinkles). i also tried to get this look with "skin" but it just wouldn't...

here's some general findings/tips for using "simple skin":

work in "sRGB" colorspace.
in image editor set:
8-bit color images to "sRGB" and their alpha to "linear" (if they have one)
8-bit greyscale images to "linear"
8-bit normal maps to "linear"
32-bit images (HDR) to "linear"

in "render globals" activate: raytrace shadows, transparency, reflection, refraction.
set "shading samples" to 4, "light samples" to 2 (for a less noisy base quality)

in "surface editor":
set skin surface to "double sided". this will avoid shading issues from bumps.

in "simple skin" node:
tweak the default settings as they are too intense:
diffuse:30%, epiderm visibility:60%, subderm visibility:40%, respect bump:100% (so the normal map will show)
note that the default "simple skin" diffuse color is this pale oliveish grey... this is actually how it should be!
so route your skintoneish color texture through a "color tool" node with the settings:
hue:140%, satur:30%, bright:30%, contrast:0% to get the proper diffuse color. again: the base image's skintone color is actually not directly used.

should you (like me) want to use "color layer" and "scalar layer" nodes for texturing... DON'T! as they seem to have issues with uv seams and produce shading errors in "simple skin". use DP's "color layer+" and "scalar layer+" instead. the standard image node works as it should.

Greenlaw
04-09-2013, 08:47 PM
Nice work! Thanks for posting this info.

BTW, I went back and re-read some of my posts in this thread from 'long ago' and, since that time, I've become a non-linear color space convert, so please forget the nonsense I wrote back then. :p

G.

Quantenschaum
04-10-2013, 08:55 AM
Nice work! Thanks for posting this info.

BTW, I went back and re-read some of my posts in this thread from 'long ago' and, since that time, I've become a non-linear color space convert, so please forget the nonsense I wrote back then. :p

G.

hey, no problem! :bowdown:
the color-space in one thing, but there's lots of conflicting info about the current 3 skinshaders "Skin", "Simple Skin" and "Fast Skin" which is sometimes based on improper settings all over the place in LW. for example: an average human skin-look can really do without "Refraction" and "Simple Skin" will also F9 render correctly, but VPR needs "Refraction-ON" to display correctly. it would also intuitively seem incorrect to set the surface to Double-Sided, but my tests show that Double-Sided just works better because it eliminates many shading errors produced by bumps in conjunction with flat lights. (i lack the proper word! with "flat" i mean: "not head on"). oh... and the "Color Layer" node shading errors with "Simple Skin" led me to believe that "Simple Skin" is broken for about 2 months!

since this headscan was free, i could upload my scene here. it's a biggie because of the texture images, but since it's the best skin i was yet able to produce, i would gladly share.

djlithium
06-12-2013, 07:59 PM
113455
hi guys, i only recently switched from LW 10 to 11.5 and of course i checked the "Skin" node right away. it's a cool node, but in my opinion it's more for doing pixar-style skin or general sss surfaces. i like the roughness feature and i wish that "simple skin" had that too, but in the end "simple skin" still rules where realism is concerned.
my "simple skin" pros: it has 3 color slots giving more texturing possibilities and it actually does render in VPR too. good looking skin is mainly a product of epidermis + subdermis, with only a hint of diffuse color, that's why i find the 2 color slot approach of "Skin" somewhat limiting.

the head in my render is a highres scan which came with a 4k color, normal and a 32-bit displacement map. based on the color image i created an epidermis and subdermis image. with the "nDo normal" photoshop plugin i made a "cavity map" based on the default normal image (for pores and wrinkles). i also tried to get this look with "skin" but it just wouldn't...

here's some general findings/tips for using "simple skin":

work in "sRGB" colorspace.
in image editor set:
8-bit color images to "sRGB" and their alpha to "linear" (if they have one)
8-bit greyscale images to "linear"
8-bit normal maps to "linear"
32-bit images (HDR) to "linear"

in "render globals" activate: raytrace shadows, transparency, reflection, refraction.
set "shading samples" to 4, "light samples" to 2 (for a less noisy base quality)

in "surface editor":
set skin surface to "double sided". this will avoid shading issues from bumps.

in "simple skin" node:
tweak the default settings as they are too intense:
diffuse:30%, epiderm visibility:60%, subderm visibility:40%, respect bump:100% (so the normal map will show)
note that the default "simple skin" diffuse color is this pale oliveish grey... this is actually how it should be!
so route your skintoneish color texture through a "color tool" node with the settings:
hue:140%, satur:30%, bright:30%, contrast:0% to get the proper diffuse color. again: the base image's skintone color is actually not directly used.

should you (like me) want to use "color layer" and "scalar layer" nodes for texturing... DON'T! as they seem to have issues with uv seams and produce shading errors in "simple skin". use DP's "color layer+" and "scalar layer+" instead. the standard image node works as it should.

Can you send me the nodal flow and maps for this? I'm setting some stuff up using the basic Daz Characters at the moment and I want to get close where you got with your head scan render images.