ironsoul posted at 10:11PM Sat, 29 June 2019 - #4355394

If I make a slight adjustment to radius eg change "1,1,1" to "1,1,1.00001" it works ok even with scale set to 0.01

Cool finding, it explains why i never have been run into this problem. Reminds me somehow of the problem with texture blur at a value of exact zero...

ironsoul posted at 4:16PM Sat, 29 June 2019 - #4355394

If I make a slight adjustment to radius eg change "1,1,1" to "1,1,1.00001" it works ok even with scale set to 0.01

Huh. I wonder if this is related to the bug in the Cycles SSS, that at zero Texture Blur you get problems with nearby transparency but if you set it to 0,0001 it gets fixed.

In any case, thank you, I'll test this later!

Nagra_00_ posted at 4:24PM Sat, 29 June 2019 - #4355397

Reminds me somehow of the problem with texture blur at a value of exact zero...

LMAO same hat!

Ironsoul this is a very interesting discovery. I am experimenting some more but it may provide the workaround we need.

I'm using a simpler scene - a stack of thin bars of varying thickness and a single infinite light. In front of the stack are two large black boxes to create shadows, allowing only a thin band of light to strike the stack.

I render from the other side to see the transmission of the light, attenuated by 3,2,1.

With attenuation set to 1,1,1, light transmits will through the top four bars. This is expected.

But with the color set to cyan (0, 255, 255), having removed all the red, we see that NONE of the green or blue can transmit past the first bar. This is very wrong because if the red is on, the green and blue transmit in equal amounts as the red. They should still transmit the same as in the red=255 case.

However, if I use a cyan light with the white surface color, it works as expected.

Following ironsoul's tip, I found that changing ANY of the three RGB attenuations to 1.01 makes it work correctly. Here it is 1.01, 1, 1

So far we know that for attenuation distance set to R, G, B

Correct: R != 1 or G != 1 or B != 1 Incorrect: R = G = B = 1

Exploring the Incorrect a little more I wondered if it would be wrong with R = G = B = 2. It is.

Changing any of the three gives this instead

One more test - this time all three attenuation distances at 1.5 - INCORRECT

It would seem that we get incorrect results if and only if R = G = B and it doesn't matter what they actually are set to. They cannot be the same.

I haven't proved it yet though.

Another question - does it matter if the values SAY different numbers, or does it matter what the effective, calculated number is?

I replaced the red value 1.5 and used a math node to supply 1.5 instead. The result is still INCORRECT

Conversely if all three of the parameters SAY 1.5 but a multiplier in the green channel changes that one to slightly more, they're effectively different and the result is CORRECT.

I've done numerous other tests with more nodes as input and so far found no inconsistency. In particular, plugging in a gray map (i.e. same value in all channels) into the scatter distance inputs will trigger the bug. At least one of them has to be different from the other two

INCORRECT if and only if R = G = B

I went back to LaFemme's ear but I think I'm struggling with the fact that the ear is just too thick, too unlike a human ear.

This is looking down at the right ear, focusing on the curve where the back of the ear attaches to the head. This is really thick, completely unlike real humans where the thinnest part of the ear is close to the head.

Yea this geometry is no good. After manipulating your subdermal map to produce a distinctly visible vein that is lit (only where the white arrow points) the rest is just black and it's not the "blue bug" we're looking at. It's the geometry.

With the same material but now the whole face lit, I get this. The veins at the temple don't look right. Eyelids do though. (forgot to do the same material on the lips - I hate lip material zones - pointless)

Interesting Bagginsbill and Ironsoul. This R=G=B bug may explain some of my unexpected results-not saved. What about having two or 3 grey scale maps at slightly different contrast levels for RGB scatter distance? I'll try that when I get home later.

Note-there are several requests to fix La Femme's ears

If you want a single channel of scatter distance you plug it into all three but give them slightly different constants. Inputs are multiplied with the plugged in value.

AO - a shader with your maps. I only used 2 of them, plus the standard LF bump map.

Oh I come back to so much! Let's see.

1- ok now I'm considering to morph her ears along with her face for this character. 2- lip materials are annoying to test materials but they're important for end results - to allow users to combine makeup with lipstick colors. 3- thank you so much, I'll test the shader when I get to my computer (I'm on mobile lounging in bed, it's a lazy Sunday for me). 4- MOTION FOR RENDEROSITY TO GET BAGGINSBILL IN THE NEW POSER TEAM. Please?

(btw, probably the eyelid veins look good because they're actually visible in the texture map - the photos I used for this are of a person with very visible veins in her eyes. I painted over them for the subdermal map, but every other vein I painted just pretty much guessing what they'd look like)

This render was done just to show the ear morph. I didn't change the default lighting or do anything to the textures.

Blackhearted's HD morphs contain a morph labeled "Ear Define", which makes her ears look much better.

"lip materials are ... important for end results - to allow users to combine makeup with lipstick colors"

That's ONE way, and perhaps the easiest way for people. But don't say it's the only way, and particularly don't say it's the best looking way as it is the worst.

The lip material zone is not necessary to get lip effects separate from skin effects - a lip mask works as well and allows lips (or eyeshadow or eyeliner or lip liner or any other makeup) to be blended with any skin, any eye, any rouge etc.

You have limitations you accept on the eyes and cheeks (one map) but the community long ago seemed to think that was unavoidable. It is avoidable and is done with masks.

Once you start using masks to control makeup layering effects, and you have the freedom to independently choose each touch, you'd never go back to the highly regimented lip material zone.

bagginsbill posted at 5:23PM Sun, 30 June 2019 - #4355556

"lip materials are ... important for end results - to allow users to combine makeup with lipstick colors"

That's ONE way, and perhaps the easiest way for people. But don't say it's the only way, and particularly don't say it's the best looking way as it is the worst.

The lip material zone is not necessary to get lip effects separate from skin effects - a lip mask works as well and allows lips (or eyeshadow or eyeliner or lip liner or any other makeup) to be blended with any skin, any eye, any rouge etc.

You have limitations you accept on the eyes and cheeks (one map) but the community long ago seemed to think that was unavoidable. It is avoidable and is done with masks.

Once you start using masks to control makeup layering effects, and you have the freedom to independently choose each touch, you'd never go back to the highly regimented lip material zone.

I have to confess I've never used material layers - so I completely forget they exist.

What happens when you go through several lip colors? Will those layers stack one on top of the other or will they just replace the previous upper layer?

Oops I found a problem - it seems that Firefly doesn't understand the SeparateRGB node.

Is there any way to separate them with math nodes? Nodes that aren't from Cycles? This will be a store product - I need it compatible with Firefly and Superfly.

Would you not just create a separate firefly shader based on a regular poser root node.

When it renders in firefly it uses that shader and when it renders in superfly it works off the superfly root.

Afrodite-Ohki posted at 4:45PM Sun, 30 June 2019 - #4355562

Oops I found a problem - it seems that Firefly doesn't understand the SeparateRGB node.

Is there any way to separate them with math nodes? Nodes that aren't from Cycles? This will be a store product - I need it compatible with Firefly and Superfly.

meatSim posted at 5:47PM Sun, 30 June 2019 - #4355566

Would you not just create a separate firefly shader based on a regular poser root node.

When it renders in firefly it uses that shader and when it renders in superfly it works off the superfly root.

I've been doing that for all my products using Cycles roots - but I'm told that for some people the Cycles root takes too long to render.

As I'm avoiding to use it now because of that, I prefer to stick to just one shader for both renderers. I'd been using EZskin for the Firefly version and a Cycles thing for Superfly for my previous character, but for this one I'd rather avoid the huge amount of nodes that EZskin uses, even though it has excellent results all those nodes seem to take a pretty long time to render too. And then as I'd have to figure out a shader for Firefly - might as well still use this one, if it's possible to achieve without the SeparateRGB.

Afrodite-Ohki posted at 5:02PM Sun, 30 June 2019 - #4355567

... I'd have to figure out a shader for Firefly - might as well still use this one, if it's possible to achieve without the SeparateRGB.

Firefly can use the component node to separate red/green/blue color components. Use 0, 1, or 2 in the component value field to return value of red, green, or blue.

seachnasaigh posted at 7:19PM Sun, 30 June 2019 - #4355568

Afrodite-Ohki posted at 5:02PM Sun, 30 June 2019 - #4355567

... I'd have to figure out a shader for Firefly - might as well still use this one, if it's possible to achieve without the SeparateRGB.

Firefly can use the component node to separate red/green/blue color components. Use 0, 1, or 2 in the component value field to return value of red, green, or blue.

BEAUTIFUL. Thank you so much.

God it's so frustrating I'm so sorry I keep asking things.

BUT. I don't think that Firefly is understanding Physical Surface's ScatterDist at all.

Look (ignore the lips, I'm testing stuff and not changing its material):

Doesn't look like it's scattering at all.

Here, I tried a simple Subsurface Skin with a PoserSurface to see if it changes, and it does:

Seems I'll either have to just go with a PoserSurface after all, or with a different shader for each renderer...

I don't know if this will be helpful to anyone at this point, but Substance Painter has SSS support now, and here's a ten minute tutorial on how to paint a scatter map

Substance Painter - Subsurface Scattering

Deecey posted at 7:58PM Sun, 30 June 2019 - #4355574

I don't know if this will be helpful to anyone at this point, but Substance Painter has SSS support now, and here's a ten minute tutorial on how to paint a scatter map

Substance Painter - Subsurface Scattering

It will to me - thank you!

Afrodite-Ohki posted at 8:34PM Sun, 30 June 2019 - #4355575

Deecey posted at 7:58PM Sun, 30 June 2019 - #4355574

I don't know if this will be helpful to anyone at this point, but Substance Painter has SSS support now, and here's a ten minute tutorial on how to paint a scatter map

Substance Painter - Subsurface Scattering

It will to me - thank you!

Great. Hope it helps!

Eric Walters posted at 11:33PM Sun, 30 June 2019 - #4355529

Note-there are several requests to fix La Femme's ears

Yes, I'm aware of this.

Unfortunately, there's a catch. At this point any changes to the ears will have to be applied as a morph, and not to the base figure itself. A bug was discovered in Poser late in the game. For example, we tried to smooth out the fingers before she was released. And we exposed a bug in Poser that any change in the base OBJ caused high resolution morphs to spike in unpredictable ways. These spikes DO NOT occur when changes are made with morphs. So we had to provide finger fixes as a morph rather than changing the base.

So since ear fixes can only be applied by morphs at this point, if there are one or more souls out there that are really good at ear shapes (I am not, trust me) ... maybe a simple ear pack would be a cool thing to put out there.

Afrodite-Ohki posted at 10:12AM Mon, 01 July 2019 - #4355573

BUT. I don't think that Firefly is understanding Physical Surface's ScatterDist at all.

I heard about that limitation with FF before but never tried. I also don’t know which algorithm is used for SSS calculations with SF. There is another limitation of the PhysicalSurface node with FF, it has no support for ray-reflection.

Nagra_00_ posted at 9:47AM Mon, 01 July 2019 - #4355616

Afrodite-Ohki posted at 10:12AM Mon, 01 July 2019 - #4355573

BUT. I don't think that Firefly is understanding Physical Surface's ScatterDist at all.

I heard about that limitation with FF before but never tried. I also don’t know which algorithm is used for SSS calculations with SF. There is another limitation of the PhysicalSurface node with FF, it has no support for ray-reflection.

Sigh. Two roots in the shader it is then - I love the look of this in Superfly, so I'm keeping this shader for it. I'll try some CustomScatter for Firefly and see what I can get with it!

Deecey posted at 10:04AM Mon, 01 July 2019 - #4355595

Eric Walters posted at 11:33PM Sun, 30 June 2019 - #4355529

Note-there are several requests to fix La Femme's ears

Yes, I'm aware of this.

Unfortunately, there's a catch. At this point any changes to the ears will have to be applied as a morph, and not to the base figure itself. A bug was discovered in Poser late in the game. For example, we tried to smooth out the fingers before she was released. And we exposed a bug in Poser that any change in the base OBJ caused high resolution morphs to spike in unpredictable ways. These spikes DO NOT occur when changes are made with morphs. So we had to provide finger fixes as a morph rather than changing the base.

So since ear fixes can only be applied by morphs at this point, if there are one or more souls out there that are really good at ear shapes (I am not, trust me) ... maybe a simple ear pack would be a cool thing to put out there.

I can offer an attempt at an ear morph for her base package if you'd like, by the way.

Hi Deecey-that would be great! Afrodite-Ohki posted at 9:14AM Mon, 01 July 2019 - #4355648

Deecey posted at 10:04AM Mon, 01 July 2019 - #4355595

Eric Walters posted at 11:33PM Sun, 30 June 2019 - #4355529

Note-there are several requests to fix La Femme's ears

Yes, I'm aware of this.

Unfortunately, there's a catch. At this point any changes to the ears will have to be applied as a morph, and not to the base figure itself. A bug was discovered in Poser late in the game. For example, we tried to smooth out the fingers before she was released. And we exposed a bug in Poser that any change in the base OBJ caused high resolution morphs to spike in unpredictable ways. These spikes DO NOT occur when changes are made with morphs. So we had to provide finger fixes as a morph rather than changing the base.

So since ear fixes can only be applied by morphs at this point, if there are one or more souls out there that are really good at ear shapes (I am not, trust me) ... maybe a simple ear pack would be a cool thing to put out there.

I can offer an attempt at an ear morph for her base package if you'd like, by the way.

How's this looking?

Before/after

Too thin in some places, I guess. I'll adjust.

Better! Is it too thin compared to a real ear? Afrodite-Ohki posted at 11:52AM Mon, 01 July 2019 - #4355673

How's this looking?

Before/after

Too thin in some places, I guess. I'll adjust.

Here is V4 using the Physical Mat. I upped the scatter distance to 4 mm. A displacement map for eye wrinkles (SubD level 3).I am also using a single light and the Envirosphere for lighting.

Eric Walters posted at 3:14PM Mon, 01 July 2019 - #4355677

Better! Is it too thin compared to a real ear?

I'm not sure, everything I do is based on "it looks right/wrong", and I think there's too much light coming through in those spots so they feel too thin for me.

Afrodite-Ohki- sounds like a reasonable solution. I realize that not everyone has Poser11. I have never liked Firefly (although I go all the way back to Poser2 and 3 with the plastic look-so I appreciated the advancement!). I TRY and go for "realistic" when I can in my renders. I used to export Poser characters to LightWave (and spent many many hours trying to make them work with SSS in LW. Having ray tracing and SSS in Poser has been wonderful! Afrodite-Ohki posted at 12:24PM Mon, 01 July 2019 - #4355645

Nagra_00_ posted at 9:47AM Mon, 01 July 2019 - #4355616

Afrodite-Ohki posted at 10:12AM Mon, 01 July 2019 - #4355573

BUT. I don't think that Firefly is understanding Physical Surface's ScatterDist at all.

I heard about that limitation with FF before but never tried. I also don’t know which algorithm is used for SSS calculations with SF. There is another limitation of the PhysicalSurface node with FF, it has no support for ray-reflection.

Sigh. Two roots in the shader it is then - I love the look of this in Superfly, so I'm keeping this shader for it. I'll try some CustomScatter for Firefly and see what I can get with it!

Eric Walters posted at 4:42PM Mon, 01 July 2019 - #4355705

Afrodite-Ohki- sounds like a reasonable solution. I realize that not everyone has Poser11. I have never liked Firefly (although I go all the way back to Poser2 and 3 with the plastic look-so I appreciated the advancement!). I TRY and go for "realistic" when I can in my renders. I used to export Poser characters to LightWave (and spent many many hours trying to make them work with SSS in LW. Having ray tracing and SSS in Poser has been wonderful!

La Femme requires Poser 11, so I assume everyone buying my stuff is in P11 - but a lot of people are still on Firefly, mostly because Superfly takes too long to render for them.

I only use Superfly too - the only reason I've been touching Firefly lately is just to test if the materials are working properly for the stuff I make.

Comparison of Physical (4mm Scatter distance) Top and Cycles (from EzSkin3-bottom. No lighting changes-just switched shaders.

Interesting comparison, The difference in specular response could be the physical surface node applying limits. Experiments that measured the F0 specular response of skin put it around 4% to 7%. Possibly the physical surface node could also be smoothing the highlights due to SSS blurring but I haven't tested that idea. It's been suggested that natural skin oils will have a blurring effect so the lack of super sharp highlights is not necessarily a bad thing.

Just to add, I'm just assuming the default spec value in the physical surface node is at the lower end of around 4% due to SM experimenting with Substance Painter which defaults to that value.