Quote -  I'm confused- did Danae acknowledge BB or not?

Before this thread, no, but she added it and wrote to me. It was an oversight, not intentional.

I, too, am sure that baby skin is as translucent as we get with scale=1. I think the range 1.25 to 1.75 is reasonable.

Quote - > Quote - ... I'm confused- did Danae acknowledge BB or not?...

...

She did acknowledge Bagginsbill's work. Very nice of her, too! I was wrong.

Weren't wrong. She read this thread and fixed it.

Scale here is 1.25. Shader is my S+B shader, not the one it comes with.

Lit with one infinite and using IDL with EnvSphere.

I used the red lipstick texture to avoid the burned-in specular problem.

Quote - I'm working on other stuff, but doing little test renders in the background. Thought I'd share.

Scale here is 1.25. Shader is my S+B shader, not the one it comes with.

Lit with one infinite and using IDL with EnvSphere.

I used the red lipstick texture to avoid the burned-in specular problem.

Danae made some incredible skin. Nicely enhanced with your shader...

Speaking of burnt-in spec... struggling with that atm... when you get a chance. In case you missed it. Being a bit thick about this, I think... :blink:

bb did you paint the wispy hair in post?

If not, how achieved?

Quote - Speaking of burnt-in spec... struggling with that atm... when you get a chance. In case you missed it. Being a bit thick about this, I think... :blink:

I saw that already but I'm really busy with work. Doing an all nighter now, coding for cash!

I did some experiments with a new tactic for removing highlights. It's not awesome yet, but it's interesting. Nothing worth publishing yet.

Quote - bb did you paint the wispy hair in post?

If not, how achieved?

Not painted. I bought a hair set with my gift certificate at the same time that I bought Dublin.

It's this set

http://www.renderosity.com/mod/bcs/sissy-le-beaux-hair-sets/87741

You get two hairs for under $10 and they are pretty good.

Quote - > Quote - Speaking of burnt-in spec... struggling with that atm... when you get a chance. In case you missed it. Being a bit thick about this, I think... :blink:

I saw that already but I'm really busy with work. Doing an all nighter now, coding for cash!

I did some experiments with a new tactic for removing highlights. It's not awesome yet, but it's interesting. Nothing worth publishing yet.

Not a worry! Good luck with it!

I used the gold lame' shader I published years ago. It looks even better now with IDL + GC.

The EnvSphere is a free HDR called doge something or other.

that is saint marks square, venice, i believe

pose tip: with inverse k on, grab hips, move around, or use hip rotate and translate dials

Really hard to code and pose all at the same time. I know, I tried, and I'm only writing in Python....

Quote - I did some experiments with a new tactic for removing highlights. It's not awesome yet, but it's interesting. Nothing worth publishing yet.

This sounds interesting to hear.  Particularly how you would end up with the right color after you've figured out where the burnt in specular and shadows are.

Here's the BiS idea I'm noodling around.

Suppose you have a pretty strong (but obviously not exact) knowledge of the ratio between red (r) and blue (b) in a given area of a texture map. For skin that is pretty smooth, or for lips, both are generally consistent. Some pixels are brighter or darker, but still the same hue, and this relationship is basically expressed like this:

r = kb

This is saying the red is some known, fixed multiple of the blue (approximately).

We could also say something about green, but I'm not that far along yet.

OK, so what does specularity do? It adds something to red, green, and blue in roughly equal amounts. So let's call the amount of specular highlight (which we don't know and varies widely) the letter s.

And let us denote the actual color components that we find in a texture with the capital letters R, G, or B. (Again I'm ignoring green for the moment so I'm only going to use equations for red and blue.)

Now, the true red value, little r, relates to the texture red value, big R, as follows:

R = r + s

Similarly:

B = b + s

Why? Because that is how highlights work. The diffuse r, g, or b value gets added with the specular highlight value (gray/white).

So our problem is how to get back the true diffuse value r, g, or b?

Well, I'm assuming that for generally pink skin, we know the value of k, such that:

r = kb

Let's replace r with kb in the first equation. The two equations are then

R = kb + s

B = b + s

Let's multiply the second by k.

kB = kb + ks

And now subtract the first equation:

kB - R = kb + ks - kb - s

This simplifies to:

kB - R = (k - 1)s

Solve for s.

s = (kB - R) / (k - 1)

Interesting. We can deduce the amount of specular highlight by doing a weighted sum of the texture value of red and blue, B and R. I was doing this in the old post from long ago, but not with this kind of precision.

So what would we do with that?

Recall the original is the sum of diffuse and specular. Well - just remove the specular!

R = r + s

R - (kB-R)/(k-1) = r

And there is the diffuse value, r, isolated!!

We would do the same for green and blue.

When k is consistent (i.e. when the hue and saturation of the true diffuse color are consistent) then this works perfectly. When hue or saturation changes it does not work perfectly, but if the hue or saturation changes are small, it works well.

I have not had time to really see how it works in practice.

Quote - Well, I'm assuming that for generally pink skin, we know the value of k, such that:
r = kb

k?

Ratio?

This is saying the red is some known, fixed multiple of the blue (approximately).

That multiple needs a name so we can work with it. I called it k.

It's a ratio because

k = r/b

Mmm.

If the extra light screws with the ratio you might have a problem.

Imagine an RGB of 250,150,200.  If the light adds 30 to each value the R portion might be a problem because it'd be capped at 255... your final ratio would be 255,180,230 and I'm not sure what the extra 25 from the red would be doing to the other two.  Now if it instead adds 1/7th of the way to 255 from the base to each... not a problem.  Not sure how this works.

Yay! I so got something.

k is the ratio! As is that 4x - 1.39y equation. Somehow you figured that the red to blue ratio was 4 to 1.39. In that previous thingie. Yes?

If so, things are starting to gel, slightly. Still looking for a Photoshop thingie that tells you what percetage of your image is red. ...or... Ish.

Now, we're moving rather quickly... perhaps a bit too quickly for these old grey cells.

Quote - Well, I'm assuming that for generally pink skin, we know the value of k, such that:
r = kb

This bit... okay. The red component = blue * the ratio of red/blue

Quote - Let's replace r with kb in the first equation. The two equations are then
R = kb + s
B = b + s (true red value, little r, relates to the texture red value, big R)

Following you so far... in English:
"texture red value, big R" = the ratio of red/blue * blue + specular highlight.
"texture blue value, big B" = blue + specular highlight

Quote - Let's multiply the second by k.
kB = kb + ks

...so red/blue ratio * texture Blue value Blue = red/blue ratio * blue component + red/blue ratio * specular highlight.

Yes? If I've deciphered this correctly, then ... what does this do / why? This need clarifying before we go to:

Quote - And now subtract the first equation:
kB - R = kb + ks - kb - s

It doesn't do anything. It has no meaning. It's just algebra. Intermediate versions of an equation are just rearrangments on the way to the goal. They don't necessarily mean anything, although sometimes they do.

The goal is to isolate s.

Do you know what a "roux" is? Nobody eats the roux, but it's the first thing you make on your way to an edible dish in a lot of interesting dishes.

Intermediate equation. Gotcha. This is where I generally get lost: the path. I'll just accept the thing as stated and see how this translates to matmatic... and a node set.

[click for full]

Rare warm brilliant day on the Irish seacoast.

lit by Mec4D (Cath) "Sunny Day" Light Probe.

::::: Opera :::::

[clik for full]

sunny

::::: Opera ::::

Very interesting thread.  With your face morph, OG, the model looks remarkably like my grand niece.

just thought i'd post a bw.....

::::: Opera :::::

[click for full]

This is V4 with the BB SSS shader via danae. I swaped out the texture map. It is the Lana map from Daz. Light is the "Sunny Day" probe from Mec4D with only one other light for specular.

No postwork. I know there is an issue with the location of the eyelashes, just no time or drive to fix that right now.

Folks, we've got us a pretty powerful little program here for this kind of work.

::::: Opera :::::

P.S. Am I supposed to be using some sort of dome or sphere? I have indirect light turned off.

[click for full]

I like to do things like this in postprocessing, push the brightness and contrast until you get a stylization of the render. In this case, the depth from the texture and the shader are still evident. That conveys some power.

:: og ::

The original pitures of the Dublin character are great, but I detect some excess blue around and over the mouth. Is this caused by the low scale in the scatter node? Is it too much translucence in the shader? How would you remove it? I see the other shaders used to modify the original don't seem to have the blue look either.

What do you think is causing this and does andyone else see it?

Mike

If you have sss in occluded mesh shaders (for instance, teeth and gums and such) you risk problems per the docs that came out with the SR1 for P9/PP2012.

According to Bagginsbill, optimal settings for skin include setting scale for scatter at 1.75. If your figure doesn't display that blue cast, you can go lower. That blue cast around the lips appears to be an arbitrary thing: some of my figures display it with scale set at 1, others don't. Mariko is notorious for having it show to some degree even with scale set at 1.75.

Go figure. :glare:

I know this because red + blue equaled black (red flag) with color add and green plus blue equaled a similar shade of green.  Without knowing any better, it's like the blue has a minimal value of blue and a negitive value of red.  Use the unattached subtract nodes to see what I mean.  I thought I had the math right, but appearently not.

Looking at this shader of yours, Winterclaw... not really sure what I'm looking at, though. Did you generate this with matmatic?

I am at a client site and can't help till later.

I made a cleaned up image to make it easier to follow allong.  Color Math 5 and 10 (R and B) are their to to keep only the red and the blue parts respectively of the color map.  Math functions 3 at the bottom (labeled k) is the ratio of red to blue.  r and b (color math and CM6) are supposed to be the post specular values.  r works okay for me, but b doesn't. 

Note green is to the far right of the mat file and follows the same set up.  The problem is that color math 6, b, doesn't combine correctly with R or G. 

Ok, I just had a thought but I don't know if it'll work.  Assuming I set the nodes up correctly and I really do have R and r, by definition I can create s (specularity from them).  If I used a math function subtract to remove r from R, I should have a mask that will work with blue and green.  Then all I'd have to do is subtract s from B and G without having to create an individual set up for all 3.

Fatal mistake - color math.

This is not color math. You want to use a Component node to pull the red channel and another to pull the blue channel - both will give you numbers. Then you do the math with the numbers.

Color math is equivalent to three independent copies of ordinary math, operating on separate data streams. This is why you can't get the red and blue and green to interact - they are in different parts.

Go to the thread Robyn created to discuss despec'ing and look at the original version I posted in 2007. Start from there.

So Robyn, you're saying that the blue cast by the mouth is a byproduct of SR1 and the renders on the vendors page that do not have that blue must  have been done in the original Poser 2012? Too strange. how do you recomend fixing or adjusting for this?

I noticed another thread about differences in renders with SR1...somethings happening here but I don't know what it is.

Mike

@ mikegg

No, The blue cast around the mouth is NOT from SR1.

Those bue area's, mostly around eyes and mouth occur when 2 different SSS surfaces are too close together.
It is in the origional PP2012 and in the SR1.

If, and I say, IF you get them, the trick is to only put SSS on the visible surfaces and not on surfaces laying behind.
Or?? Change the camera angle 5° to 10°, mostly they are gone.

Sometimes they are even visible when SSS is on the skin, and SSS is on the clothing.  
Again turn the camera 5° to 10°, and they are gone.

The blue area's become visible under certain combinations of IDL, SSS calculations; and camera angle when 2 SSS surfaces are too close together.

Example: If the mouth is closed, do NOT use SSS on the inner mouth parts.

Nothing NEW.
Nothing SR1 generated, it has always been there.

The confusion about SR1 is this - Smith Micro did not publicly acknowledge the blue-skin problem until the release of SR1. In the updated release, they noted it for the first time.

I noted it in beta before the initial release, but I guess they hoped it would be only a minor problem that I ran into because I'm a guy who pushes Poser rather hard.

But it's coming up in ordinary situations and there isn't much you can do about it other than to note that this is why unbiased renderers are great. Because they have no cheats and approximations, you don't find situations where the cheating reveals itself as a flaw or artifact.

I would love it if the community would embrace an unbiased render engine built into Poser.

while i believe I see the "blue effect" in my render just above on this page...

Posted Tue, Jan 3, 2012 11:39 pm

...are you guys seeing it in my render in the OP and the other initial posts on page 1?

::::: Opera :::::

Thanks for the clarification from both of you. I misunderstood, but I knew it didn't look right. It could be a pain to go to each internal gums teeth mouth sufrace, but now I understand the problem better. It is a beautiful character and I guess one just has to compensate for the blue tone. Perhaps because of the fair skin tone it is more pronounced.?

thanks again, Mike

Yes I see it in the first post.

Quote - The confusion about SR1 is this - Smith Micro did not publicly acknowledge the blue-skin problem until the release of SR1. In the updated release, they noted it for the first time.

I noted it in beta before the initial release, but I guess they hoped it would be only a minor problem that I ran into because I'm a guy who pushes Poser rather hard.

But it's coming up in ordinary situations and there isn't much you can do about it other than to note that this is why unbiased renderers are great. Because they have no cheats and approximations, you don't find situations where the cheating reveals itself as a flaw or artifact.

I would love it if the community would embrace an unbiased render engine built into Poser.

I would definitely embrace being able to either chose an external / optional renderer or vote for another one to Firefly, personally. But I understand there are mat-room considerations. Materials would have to be tailored for the new renderer... unless I'm missing something. IOW, a major re-write.

Quote - Intermediate equation. Gotcha. This is where I generally get lost: the path. I'll just accept the thing as stated and see how this translates to matmatic... and a node set.

I posted the matmatic version. You can publish the mt5 for other folk if you like. Gotta go.

http://www.renderosity.com/mod/forumpro/showthread.php?thread_id=2841945&page=2

It would be great! I'm not sure how bad the mat situation would be- I see a lot of Luxrender scenes that look great! I've not tried it myself

Quote - > Quote - The confusion about SR1 is this - Smith Micro did not publicly acknowledge the blue-skin problem until the release of SR1. In the updated release, they noted it for the first time.

I noted it in beta before the initial release, but I guess they hoped it would be only a minor problem that I ran into because I'm a guy who pushes Poser rather hard.

But it's coming up in ordinary situations and there isn't much you can do about it other than to note that this is why unbiased renderers are great. Because they have no cheats and approximations, you don't find situations where the cheating reveals itself as a flaw or artifact.

I would love it if the community would embrace an unbiased render engine built into Poser.

I would definitely embrace being able to either chose an external / optional renderer or vote for another one to Firefly, personally. But I understand there are mat-room considerations. Materials would have to be tailored for the new renderer... unless I'm missing something. IOW, a major re-write.

Quote - It would be great! I'm not sure how bad the mat situation would be- I see a lot of Luxrender scenes that look great! I've not tried it myself

 

Quote - > Quote - The confusion about SR1 is this - Smith Micro did not publicly acknowledge the blue-skin problem until the release of SR1. In the updated release, they noted it for the first time.

I noted it in beta before the initial release, but I guess they hoped it would be only a minor problem that I ran into because I'm a guy who pushes Poser rather hard.

But it's coming up in ordinary situations and there isn't much you can do about it other than to note that this is why unbiased renderers are great. Because they have no cheats and approximations, you don't find situations where the cheating reveals itself as a flaw or artifact.

I would love it if the community would embrace an unbiased render engine built into Poser.

I would definitely embrace being able to either chose an external / optional renderer or vote for another one to Firefly, personally. But I understand there are mat-room considerations. Materials would have to be tailored for the new renderer... unless I'm missing something. IOW, a major re-write.

Hi Eric... the materials for a Poser scene have to be re-assigned for LuxRender ... like done in Pose2Lux. At this point, anything involving much more than an imageMap has to be redone for Lux. Pose2Lux makes this process reasonably straightforward.

in the initial renders I posted, BB is saying he sees the blue effect around the mouth. I believe I also see it around the eys. Do the rest of you agree?

Now that would be consistent with the post above that says "SSS on adjacent groups" etc., because I bet there is SSS on the "eye_socket" and on the gums/teeth.

I'll attempt a control/experiement post over the weekend.

::::: Opera :::::

I see it on the eyes too.

I can tell you this as well - I tested embedded an object inside a poser sphere and then running scatter on the sphere. I could not tell where the object was inside. That confirmed what the original paper on this scatter technique said - that no interior geometry would influence the scatter effect.

Apparently, my mistaken assumption was that the test would work the same if the interior object was scatter or not.

What we're seeing now is the interior object or mesh, if it also uses scatter, accidentally influences the scatter of nearby meshes. In a very bad way.

This is the shader that comes with a V4 char over at RNDA.  No GC or SSS in this image.

The textures on your second image "pop" more, I think. The first one is a bit more doll-like.