I have a little more testing to do, but I think that about half of this shader is doing nothing noticeable. I'll demonstrate later. Which means we can eliminate a lot of it.

There are important things it is not doing, and to get those you have to add some specific things that are found in my VSS shader. At which point, you'd end up with my VSS shader, so I'm not sure there's any point in instructing you how to transform the IS shader to the next step if, in the end, it is exactly the same as my VSS shader. :)

Quote - If you're using a texture set where the author, being lazy and/or naive, has actually made the bump map by converting the color map in any way, then you should not use that. You'd be better off with just a turbulence node.

In any way? Why? I don't see that. It seems to me that most of the skin surface is going to correspond to bump, in the sense that darker areas are probably going to be slightly recessed (pores etc). And the rest can then be corrected manually (e.g. brows, lips, nipples, moles etc) in the converted map.

No?

Show me a color map where the darker skin splotches (freckles and other blemishes) should be depressed. Or veins or capillaries.

The bump map should have the tiny grooves from the smallest wrinkles. These do not correspond to color changes.

Quote - The bump map should have the tiny grooves from the smallest wrinkles. These do not correspond to color changes.

Sure, to some degree - although tiny wrinkles do tend to be a bit darker than surrounding skin if you're using photos - but all those things are going to have to be added anyway. I guess what I don't see is why you'd want to create everything (including the pores - which yeah, you can fake with some sort of noise, but my assumption is that the photos are going to provide more authentic noise than an algorithm) from scratch, and by retaining the colour map's essential features you always can readily see where you're at on the texture. Maybe that's only important to me cause I'm a rookie. :)

Here's how I've been approaching it. Maybe it's all wrong, though it seems to be working okay for me so far.

1. Save off a copy of colour map, desaturate, and get contrast and brightness where I want it.

2. With white brush (varying opacities): Neutralize spots that are dark that shouldn't be - e.g. nipples, lips, eyebrows. (Actually, in these cases, I'm going a lot further than neutralizing, as I want them visually raised, not simply level with the surrounding skin.)

3. With black brush: Draw in wrinkle lines etc. and accentuate recesses/indentations not adequately provided contrast by the colour map. (Here again to me is the advantage of the colour map - it's easy to accentuate a surface your drawing on directly.)

All my work is on separate layers so that I can adjust the relative strength (via layer opacities, in particular) of everything until I'm happy.

Now, of course, if you're going to have a lot of freckles, it's best to add those to the colour map after you've already saved off the bump map, so neutralization is unnecessary. (Natch, if pasting from reference photos with tons of freckles, something different is probably going to be required by the approach.)

As you can see, this is a lot different than just plugging a colour map into some nodes and calling it a bump map - I've got a lot of time invested in manual tweaking and adjusting.

The thing is, before I started, I looked for some good tutorials for creating bumps... but I ended up pretty much developing my own workflow, because there really didn't seem to much available saying what the best practices would be. Mostly what I read gave the principles of how bump maps work (which is relatively straightforward, ISTM), not how best to create them step by step.

As I said, this method seems to be working for me, but if there's a better way (better results, quicker process), by all means, I'm happy to learn!

Here's what I do know. Bump maps from some "very good" Poser texture makers look terrible to me. When I read about the work of movie makers or any of the really successful "realism" artists at CGTalk, there is always a completely hand-drawn texture set involved. None of them start with a color photo and try to derive a bump map from that.

Here's an example. This is Rebel Mommy's Kyrsten. She's a very good texture artist when it comes to color, but her bump map is derived from the color map and then she does some processing and some hand-touch up.

If you click to full size and look closely, you'll see that on her left cheek the pores look indented and on her right cheek they look like they stick out. This is because the original photo of the real human had the light coming from one side.

When light hits a pore at anything but straight, one side is brighter than the other. When you use that luminance for a bump map, you get one side high and the other side low. This can cause all sorts of confusion because its the middle of the pore that should be low, not the sides.

Also look at her eyebrows. Totally messed up - because the luminance of the overlapping hairs is identical, and results in a completely ridiculous appearance.

Look also at the figure's right cheek, in the middle of the patch of pores near the nose. There is a smudge. RM must have pressed her mouse there by accident and didn't notice it.

Another problem is that the luminance of the lip is pretty uniform. Deriving the bump for the lip from the luminance results in flat lips. I see this all the time. Lips have much bigger rise and fall than any other part of the face, but you hardly ever see the lip bump excursions being bigger than the rest of the face. Usually they're less as in RM's texture set.

It boggles my mind that nobody notices the lips are flat. Or worse, they're grooved in some way that corresponds to the way light and shadow hit the lip in the original photo. Just as with pores, the lip grooves should be centered between the highlight and the shadow in the photo. And, of course, that highlight and shadow shouldn't be in the color map at all.

(Note: Rendo somehow posted this 4 times in partial edit. I had to delete the extra copies and re-edit it.)

Ouch, that bump does look bad. I don't think I'd use it, though as you said, RM is an excellent texture artist when it comes to colour maps.

Did you exaggerate the bump level here for illustration? The eyebrows look especially odd if that's the default bump amount.

Personally, I get my photos from 3d.sk, and they seem to be pretty well (evenly) lit, so hopefully I won't have those degree of issues... not noticing anything like that yet, at any rate. Unless that was the matter I was reporting on earlier, but I don't think so, as moving the lights around didn't seem to correct/hide that particular issue... I'm still thinking that probably had something to do with the way Poser saves material collections when you don't rename.

Hmm.

There is no such thing as an appropriate default bump amount. It depends on the dynamic range that was used in the bump map. There are no standards for how far the bump map should represent a given change with each change in luminance. I actually decreased the strength from the default VSS value.

When I used RM's shader instead of mine, it was the same bumpiness. It just didn't look as good because her shader doesn't have GC in it. It does have fake SSS, though, which is cool.

I'm doing another render for you. I think you'll be surprised.

I used a Turbulence node to make tiny wrinkles that break up the specular so she doesn't look like plastic. That's all. The pores look right. The eyebrows look right. Pores are not as indented as everybody thinks. Maybe a thousandth of an inch.

I can do the lip ridges procedurally, too.

Quote - There is no such thing as an appropriate default bump amount. It depends on the dynamic range that was used in the bump map. There are no standards for how far the bump map should represent a given change with each change in luminance. 

No, I realize that. By "default," I meant the amount of bump that RM herself had specified in her MAT pose (or whatever she used).

FWIW, this is what my current bump map looks like. I dunno, maybe it doesn't look colour-map derived anymore; there's a lot of handwork in it. I can see that my approach to whitening gets rid of some of the texture detail in the cheeks... I have an idea how I can get that back, though.

this is IMO ocmpletely wrong.
bump maps are used for small parts. pores,wrinkles, ......

the dark parts are ok because they will go inside. but the bright parts? 

The bright parts = outward. How is that wrong?

because you will change the shape of her face. again bump maps are only for small parts (pores,wrinkles).

you have bright cheeks and lips on the bump map. this means that it will change the lip and cheek.

Well, you would know more than I would. But I've been doing dozens of renders of this character, and I can't say I'm seeing a change in the shape of the face. The bump is simply accentuating the natural contours, is what it looks like to me... although I may well have the contour aspects too strong, I dunno.

bump and specular map tutorial

Thanks! a glance suggests that's a lot more detailed than anything I was able to find. I'm gonna have to browse around CG Society again... it's been a while.

hm, not as much to that as I first thought... will keep looking, and working....

this link was so that you understand what a skin bump is used for.

its for DETAILS on the SKIN. lips are not details.

if you want to change the face then you should use a displacement map.

I see. I had seen a number of articles and brief tutorials that employed bump maps on a more general level than that micro level of detail - here's an example; I saw a number of them: http://wiki.blender.org/index.php/Doc:Manual/Textures/Maps/Bump_and_Normal_Maps

Truthfully, keeping things to a microscopic level seems simpler and easier.

i bump can be used for everything. but when using on skin its used for details.

Okay, now that I finally have some time on my hands to go back to the drawing board, I've made what I hope to be significant improvements.  Since the original shader, however inaccurate, does seem to provide decent results, I'm going to leave it up and just release a YARST version 2.0 tutorial as soon as I perfect the latest incarnation.

Quote - At which point, you'd end up with my VSS shader, so I'm not sure there's any point in instructing you how to transform the IS shader to the next step if, in the end, it is exactly the same as my VSS shader. :)

Actually, I prefer to think of Mike as the proverbial cat that we're both trying to skin :)

Anyway, here's a sample shot of what I've done.  Further comments are welcome, as always.

www.hsmespanol.com/NewTom2.jpg

What can be seen looks good, but frankly the shot is too dark to determine a whole lot, especially since it's not a portrait shot.

IMO.

Quote - Here the figures are back-lit. This is rim lighting, although not the most extreme case.

Compare the VSS shader to yours. Because you're not using Blinn, but a Specular node instead, you're not getting the sharp bright specular on the edges. Even with the correct Fresnel (in the middle) its not there, because the Specular node just can't generate that kind of response. The Specular node is a left-over from 1985 and should not have even been included in Poser, in my opinion.

i noticed an old post from you BB.

so you think we should replace the specular node with this? this is from VSS skin shader.

should we only use the specular node for some tricks in  the shader? 

I would suggest to always use the Blinn node, yes. Use the Specular node only for tricks, yes.

Use the VSS parmetric setup? No. That was a mistake - I'm still learning, too. I have a better understanding of the SpecularRollOff now, and how I used it there is not optimal. It only works as I expect for a limited range of PM:Shine values.

interesting .

your new setup is similar to this? with edge_blend?  that was pretty awesome. :) 

Nope - not that. I did that because you wanted to know how to do the tight/front loose/back specular effect. 

aha ok.

but still the blinn from VSS is still very good. very nice how you connected the nodes so that when you change the shine you change eccentricity and specularrolloff

p.s.
this site  has some nice info on the nodes. but the POW is not explained.
www.castleposer.co.uk/my_tutorials.html

do you maybe know for a site on the internet where it is explained with images? you are using it a lot of times but i dont know what it does. the manual has nothing.

POW is Power - exponentiation, as in 2 to the 3rd power is 8, .5 to the power 2 is .25

aha. what do we connect in value 1 and 2?

That isn't obvious? Hmm. Math and Color_Math nodes all follow the same pattern. Functions that take only one value use Value_1. Functions that take two values use Value_1 and Value_2, with Value_2 being the second value.

Add is: Value_1 + Value_2
Sub is: Value_1 - Value_2
Mul is: Value_1 * Value_2
Div is: Value_1 / Value_2
**Pow is: Value_1 ** Value_2
**Sine is: sin(Value_1)
Cosine is: cosine(Value_1)

etc.

thanks. where could we use POW? where would you use it