PDA

View Full Version : SKIN shader / SUBSURFACE SCATTERING



MAUROCOR
01-06-2018, 12:26 PM
Hello you guys. Hope you are having a lot o fun playing with LW 2018.

I donīt know if someone asked about this before - lots of threads to read, sorry - but I would like to know if you are having problems with the new SKIN shader.
I just canīt make it right using NORMAL MAP.
Are you getting good results?
If yes, can you post a sample, please?

I would like to ask to the beta testers the same, examples and scenes.

There is a beautiful render that Rebel Hill did. Would be possible to share this scene once it is a free model, please?

http://static.lightwave3d.com/marketing/lightwave_2018/release/index.html#images-15

gar26lw
01-06-2018, 05:41 PM
if you are using a normal map from 2015, invert the y as (thbomk) 2018 now does things the standard way as other apps. see if that helps.

Carm3D
01-06-2018, 07:40 PM
I was having the same problem. I learned today that normal maps do nothing with the skin material. Use Principled BSDF material with subsurface settings instead. The more subsurface scattering you use, the less your normal maps will show.

gar26lw
01-06-2018, 08:36 PM
principal shader seems to work ok with normals. you could try putting normal into the surface input instead?

http://forums.newtek.com/attachment.php?attachmentid=139254&d=1515296023

139254

jwiede
01-06-2018, 08:47 PM
The more subsurface scattering you use, the less your normal maps will show.

That's not so surprising if you think about how SSS (and similar) shaders and normal maps actually work.

On SSS-heavy surfaces, the usual answer would be to switch from fine-detail normal maps to micropoly displacement, but...(shrug).

Carm3D
01-06-2018, 09:41 PM
The exception to that rule is that reflections on SSS materials are still 100% influenced by normal mapping. Just not the diffuse.

MAUROCOR
01-07-2018, 05:22 AM
Thank you for your help, guys, really appreciated!

I just found out the problem in my case.

In the EDIT IMAGE the Color Space of my Normal Map image was as DEFAULT. In LW2018 the default is sRGB. So, to make it works properly I had to change the Color Space to LINEAR and it worked like a charm.

Thank you again, guys!

Tobian
01-07-2018, 05:28 AM
The other way round Carm... Normal maps don't do much with SSS, they work for everything else... Skin consists of 3 layers of SSS and 2 spec, so they only work with the spec layers...

DogBoy
01-07-2018, 06:47 AM
The other way round Carm... Normal maps don't do much with SSS, they work for everything else... Skin consists of 3 layers of SSS and 2 spec, so they only work with the spec layers...

That's what he said: It only works on the Spec

MAUROCOR
01-10-2018, 12:45 PM
I donīt agree with the fact Normal Map works only with spec layers. There is no reazon for that. I understand if you have a great amount of SSS of course it will wash out the micro details but if not it shouldnīt do that.

Rebell Hill, I am trying to thank you for your feedback but your inbox private message is full. So thank you anyway.

Could you share your scene with us, please? There is not much sample/examples of SKIN in this release in my opinion, so if you could do that would be amazing. http://static.lightwave3d.com/marketing/lightwave_2018/release/index.html#images-15

For some reazon Antti didnīt answer my message.

erikals
01-10-2018, 04:30 PM
Could you share your scene with us, please?
personally, i think that's a bit much to ask.


For some reazon Antti didnīt answer my message.
i think that guy is superbusy.

even though it is in LW2015, Chris Jones' older posts might help a bit...
here, a Sigma2 process >
http://forums.newtek.com/showthread.php?137719-Human-Progress/page11

gar26lw
01-10-2018, 04:54 PM
Could you share your scene with us, please? There is not much sample/examples of SKIN in this release in my opinion, so if you could do that would be amazing. http://static.lightwave3d.com/marketing/lightwave_2018/release/index.html#images-15

For some reazon Antti didnīt answer my message.

were you referring to the scene for the image I posted?

Edit, ok I see, the one on the LW install

MAUROCOR
01-10-2018, 06:10 PM
A - personally, i think that's a bit much to ask.

B- i think that guy is superbusy.

C - even though it is in LW2015, Chris Jones' older posts might help a bit...
here, a Sigma2 process >
http://forums.newtek.com/showthread.php?137719-Human-Progress/page11

A - Why? He was a beta tester and the head object is free. What is the problem?

B - Yes, maybe. But I think the content is very poor for a release of this magnitude. I miss Lino here in the forum.

C - It is a different material. I would like more examples using SKIN in human models.

But thanks for your feedback, erikals. I always watch your videos and learn a lot with them. ;)

MAUROCOR
01-10-2018, 06:13 PM
were you referring to the scene for the image I posted?

If you can. Everything that could help is welcome.

erikals
01-10-2018, 06:24 PM
A - Why? He was a beta tester and the head object is free. What is the problem?
just saying that he might want to keep it to himself for misc reasons. (business/personal)
let's see though, maybe he'll share some info.

B - Yes, maybe. But I think the content is very poor for a release of this magnitude.
only running trial, so can't say what the content lacks (i think there was a 2GB content file in the full version)

I miss Lino here in the forum.
i don't think he was very active at the forums? more on LightWiki.

C - It is a different material. I would like more examples using SKIN in human models.
yes, Sigma2 is the way to go for realism though, according to Chris and Andrew.

But thanks for your feedback, erikals. I always watch your videos and learn a lot with them.
cool, i'm always happy to hear when they are of good use. thank you. :) :)

erikals
01-10-2018, 07:05 PM
edit; looks like simple skin can do a great job also,
https://docs.lightwave3d.com/display/LW2018/Skin

however, important, read Chris notes on Skin vs Sigma2
here > http://forums.newtek.com/showthread.php?137719-Human-Progress&p=1381810&viewfull=1#post1381810

(not sure how 2015skin vs 2018skin differs)

BeeVee
01-11-2018, 05:35 AM
2018 skin is a different beast, and replaces the three Skin nodes present in 2015 (like it says on the url you chose). Chris doesn't have a sample for me of 2018 skin at time of writing, but obviously I'd rather have that.

B

MAUROCOR
01-11-2018, 06:14 AM
2018 skin is a different beast, and replaces the three Skin nodes present in 2015 (like it says on the url you chose). Chris doesn't have a sample for me of 2018 skin at time of writing, but obviously I'd rather have that.

B

Yes, it is a different beast, I know. But I donīt like the fact that BUMP and NORMAL MAP is only visible with specular. It should be visible even WITHOUT specular. Of course it could be compensate if LW had MICROPOLY DISPLACEMENT, and it hasnīt - doesnīt matter if we are asking for it for 5 or 6 years now.

RebelHill
01-11-2018, 07:06 AM
Normal/bump maps are intended to simulate exceedingly small details on a surface, not larger scale ones. It's perfectly normal to expect that the diffuse element of sub surface shading would ignore these, as any meaningful scattering distance is going to dwarf the scale of such small details, and in cases where it wouldnt, then either the bump is being used to try and represent a scale its not intended for, or the SSS distance is being set so small as to have a negligible visual appearance anyway.

Try this for instance... set some kind of bump map to displace your geo (high subdivs obvs), set the displace distance to 1mm... render it with a non SSS surface, pretty lumpy. Now set a SSS surface, with the scattering distance at 2-3mm and that lumpyness is almost entirely "erased" by the scattering, even though its physically there in geo and thus being fully evaluated.

You CAN get the SSS to show bump in its diffuse, IF you set the SSS distance very small (like 1mm or less) and set the bump amp really high (like 500%+). But it doesnt look "bumpy", more "mottled".

End of all, no, bump/normal maps dont really do anything for diffuse with SSS shading, and nor should you really expect them to. If you want to use them to add detail in the diffuse portion of the shading, then the better way to do so is to use them to either tweak the surface colour itself, or, more "realistically", use them as maps to modulate the SSS distance parameter, or pehaps the diffuse weight parameter (in the multilayer skin material) thus having the amount by which lower layers are seen through the upper one vary with the bump detail.

Beyond this, it's all down to the specular picks, which respect bump and normal just fine.

MAUROCOR
01-11-2018, 07:07 PM
edit; looks like simple skin can do a great job also,
https://docs.lightwave3d.com/display/LW2018/Skin

however, important, read Chris notes on Skin vs Sigma2
here > http://forums.newtek.com/showthread.php?137719-Human-Progress&p=1381810&viewfull=1#post1381810

(not sure how 2015skin vs 2018skin differs)

I donīt understand why they used the Chris Jones picture done with LW 2015 in the LW 2018 doc. After all there is NO SIMPLE SKIN in LW 2018. What is the point?
Personally I liked a lot SIMPLE SKIN in 2015 but then you have the Fiber FX problem. Now FIBER FX is much better and I am having problems with SKIN. What the hell?!?

About using Sigma2 do create Skin, I think the 2018 version is different. I have to test yet.

- - - Updated - - -


Normal/bump maps are intended to simulate exceedingly small details on a surface, not larger scale ones. It's perfectly normal to expect that the diffuse element of sub surface shading would ignore these, as any meaningful scattering distance is going to dwarf the scale of such small details, and in cases where it wouldnt, then either the bump is being used to try and represent a scale its not intended for, or the SSS distance is being set so small as to have a negligible visual appearance anyway.

Try this for instance... set some kind of bump map to displace your geo (high subdivs obvs), set the displace distance to 1mm... render it with a non SSS surface, pretty lumpy. Now set a SSS surface, with the scattering distance at 2-3mm and that lumpyness is almost entirely "erased" by the scattering, even though its physically there in geo and thus being fully evaluated.

You CAN get the SSS to show bump in its diffuse, IF you set the SSS distance very small (like 1mm or less) and set the bump amp really high (like 500%+). But it doesnt look "bumpy", more "mottled".

End of all, no, bump/normal maps dont really do anything for diffuse with SSS shading, and nor should you really expect them to. If you want to use them to add detail in the diffuse portion of the shading, then the better way to do so is to use them to either tweak the surface colour itself, or, more "realistically", use them as maps to modulate the SSS distance parameter, or pehaps the diffuse weight parameter (in the multilayer skin material) thus having the amount by which lower layers are seen through the upper one vary with the bump detail.

Beyond this, it's all down to the specular picks, which respect bump and normal just fine.

Well, if you say. I will keep trying to improve my tests. Thank you for your feedback.

MAUROCOR
01-11-2018, 07:30 PM
I got that one but I had to add an ambient occlusion in PS to make wrinkles more visible.

https://static.wixstatic.com/media/aa2b60_e392e7f26271484793c090cca51325b9~mv2.jpg/v1/fill/w_793,h_910,al_c,q_85,usm_0.66_1.00_0.01/aa2b60_e392e7f26271484793c090cca51325b9~mv2.webp

erikals
01-11-2018, 07:41 PM
Personally I liked a lot SIMPLE SKIN in 2015 but then you have the Fiber FX problem. Now FIBER FX is much better and I am having problems with SKIN. What the hell?!?
creating believable skin is difficult, it will take some time for many of us to learn it properly using the LW2018 system.

it took quite some time to get the right settings in the LW2015 system too.
i think it's a bit early to tell.

my own tests haven't really been any better or worse with the LW2018 system.

erikals
01-11-2018, 08:03 PM
maybe ask Mikko Kunnari over at LightWiki
https://www.facebook.com/groups/lightwiki/permalink/1304950382944291

his result looks something like this >
(filmic color correction / faux tonemapping by me)

https://i.imgur.com/MkNmDbb.png


not perfect, it's a start though.

a problem tends to be that the color texture "drowns" in the sss
haven't figured out how to solve that.
subnote; Sigma2 doesn't seem to have this issue to the same extent.

BeeVee
01-12-2018, 02:33 AM
Please note that Chris used Sigma 2 for the eyeball, not the skin. I know the image is from 2015's skin, but it is such a lovely image and I also don't have a comparable one using 2018's Skin material yet. As soon as I do, the one in the Skin node documentation will be replaced with it.

B

erikals
01-12-2018, 03:25 AM
both were used, first skin, then here, Sigma2
http://forums.newtek.com/showthread.php?137719-Human-Progress&p=1381810&viewfull=1#post1381810

MAUROCOR
01-12-2018, 03:53 AM
Please note that Chris used Sigma 2 for the eyeball, not the skin. I know the image is from 2015's skin, but it is such a lovely image and I also don't have a comparable one using 2018's Skin material yet. As soon as I do, the one in the Skin node documentation will be replaced with it.

B

At the end he chose to use Sigma2 for SKIN too ( instead Simple Skin ).

http://forums.newtek.com/showthread....=1#post1381810

I thought this image was created using LW2018. http://static.lightwave3d.com/marketing/lightwave_2018/release/index.html#images-15

MAUROCOR
01-12-2018, 03:59 AM
maybe ask Mikko Kunnari over at LightWiki
https://www.facebook.com/groups/lightwiki/permalink/1304950382944291

his result looks something like this >
(filmic color correction / faux tonemapping by me)

https://i.imgur.com/MkNmDbb.png


not perfect, it's a start though.

a problem tends to be that the color texture "drowns" in the sss
haven't figured out how to solve that.
subnote; Sigma2 doesn't seem to have this issue to the same extent.

It is a good start.
Mikko Kunnari's render loses skin micro details, and micro details is what I am looking for.
Are yo using Displacement Map on it?
If so, the problem is that I am trying to figure out a way to make it work well with BUMP and NORMAL MAPS.
But them are washed out if you use high values for the Scattering channels or a low lighting.
I will keep trying.

TheLexx
01-12-2018, 10:33 AM
I do not yet have a confident handle on the new lighting and materials, but I would just say that if realistic interaction of skin in low light with micro-details is the goal, then a real film production surely wouldn't approach it in that way, precisely because the results would be so poor with lost details and noise. Presumably they would at least throw stronger light on the subject, then do some night-for-day trickery in post-production, allowing some fine details to be retained ? The eye wouldn't see micro-details in the dark, so it is a bit of an irony they harness real world interaction to fake it. Hope that makes sense and I'm not just confusing myself. :)

MAUROCOR
01-12-2018, 12:14 PM
Well, never mind. I think it is my inability to get what I want that is making me frustrated.

RebelHill
01-12-2018, 12:44 PM
I thought this image was created using LW2018. http://static.lightwave3d.com/marketing/lightwave_2018/release/index.html#images-15

It was... but again, you think you're seeing something you're not. All the tiny little details in that image are the result of either displacement (which the SSS still smoothes out quite a lot) and specular. SSS will smooth out small details, and bump/normal is intended to be so small scale that any amount of scattering would "erase" it. Bump normal do not contribute to the diffuse component of skin, nor should you expect them to.

MAUROCOR
01-12-2018, 01:01 PM
It was... but again, you think you're seeing something you're not. All the tiny little details in that image are the result of either displacement (which the SSS still smoothes out quite a lot) and specular. SSS will smooth out small details, and bump/normal is intended to be so small scale that any amount of scattering would "erase" it. Bump normal do not contribute to the diffuse component of skin, nor should you expect them to.

You made a mistake, RH. I was answering the Bee Vee post when he said: "I know the image is from 2015's skin, but it is such a lovely image and I also don't have a comparable one using 2018's Skin material yet. As soon as I do, the one in the Skin node documentation will be replaced with it." He reffered to the Chris Jones image. So I pointed out your image thatin my opinion is so beautiful as the other one.

erikals
01-12-2018, 01:18 PM
You made a mistake, RH.
i don't think so, it was a general comment, regardless of LW version.
that's how i understood it at least.

again, realistic skin does take time to achieve. there's no magic button.

erikals
01-12-2018, 01:48 PM
no time at the moment,
if you want to proceed, check out this LW2015 video >


https://www.youtube.com/watch?v=mvphqHR1UXw

the node setup >
https://scontent-arn2-1.xx.fbcdn.net/v/t31.0-8/14524345_10154576329519452_6263020658170148196_o.j pg?oh=e53f492b321a4dcc4fc6cdc41b51d1df&oe=5AFAD9F1

more FB info >
https://www.facebook.com/groups/lightwiki/permalink/910517052387628/?match=ZGF2aWQgbWFydGlu

obj files + textures >
http://ten24.info/tag/free-3d-scan/

MAUROCOR
01-12-2018, 02:20 PM
i don't think so, it was a general comment, regardless of LW version.
that's how i understood it at least.

"I know the image is from 2015's skin, but it is such a lovely image and I also don't have a comparable one using 2018's Skin material yet." - Bee Vee.

erikals
01-12-2018, 02:27 PM
we know.

i think we might be lost in translation here.

MAUROCOR
01-12-2018, 02:34 PM
I understood that this new SKIN is what it is and I will have to learn how to use it.
But I did some quick renders in LW 2015.3 only to show my point about low lights and specular.

In this example I used SIGMA2 material, 2 Lights - front and back.
Even with a low light and no specular the bump effect keeps there. It doesnīt happen using the new Skin because bumps (and normal maps) are visible with specular.

RebelHill
01-12-2018, 03:00 PM
Right... and that's the old sigma 2, which is a different material, and is more than only SSS, and ofc you can see how it still reduces the bump effect.

The new skin is SSS only, if you want something more like the old sigma 2, which lets you mix some SSS with regular diffuse shading, try either the new sigma material, or principled.

erikals
01-14-2018, 03:18 PM
partly made it...

https://i.imgur.com/2ajRAAQ.png

can't make it look nice in sRGB mode though, it just ends up looking super-flat.
no pro, so there might be a way in sRGB, but for this i had to go for Linear.

i must be doing something very wrong with the light...
nothing else to do but try harder...
an improvement though. :)