Trying to Understand Sub Surface Scattering

[markarjun]

I’m trying to get to grips with Sub Surface Scattering, but at the moment I’m at a loss on how to get it to work properly. What I cant quite understand is that the sss effect seems evident in the modelling window when rendered out in final, however not all evident in the default choreography window. I’ve been using some of Yves Possaint’s sub surface scattering value examples and testing them out on basic primitives with no luck.

I was told by Robcat that John Bigboote uses sss a lot, any help from anybody would be greatly appreciated.

[robcat2075]

I haven't done much with SSS so this is my uninformed attempt. I hope others will chime in.

 

My first gambit is to set some absurdly large settings to see if anything happens. Yes, it does!


This scene has two identical objects facing a single Sun light at identical angles. Both have identical materials on them except the right one has SSS ON with the settings shown.

 

They do indeed appear different. The light seems to flow through the second cube instead of getting cut at the corner. Notice how the shadow the first cube casts on the second one is blurry even though the Sun light is just a simple sharp-shadow light.

 

 

These settings are more like what you'd use for a waxy candle than human skin. The high values for "extinction" seem to require large values for density of SSS Samples in order to shade smoothly.

 

 

 

Here is the PRJ...

 

Marcajun_test sss_03.prj

[robcat2075]

Mark,

 

I'll also note that I shrunk the objects down quite a bit from your sample PRJ.

 

the extinction distances seem to be real distances, centimeters I presume, so the visible effect of "2" will appear to be much less on a giant object seen from a distance than "2" will make ona small object seen up close.

the cubes in my example above are about human head width.

small distances, less than 1.0, may be scientifically accurate for skin, but also are very long to render, and produce such a slight effect that they may seem not worth the time.

Larger values, greater than 1, render faster and produce a more visible scattering effect. It may be the "exaggeration" you need to get the effect you need across on the screen.

[robcat2075]

Here is a test with more conservative values.

 

The left has no SSS, the right does.

This head is about 25 cm wide. Although they are slightly different angles to the camera, they are both identical angle to the light.

 

[markarjun]

Hi Rob, many thanks for this, I was trying to use ever smaller values to try to exaggerate the SSS effect on the model little realising the opposite was needed. I also didn't factor the size of the object in relation to SSS values and its effectiveness, that does in hindsight make a lot of sense. Will continue to mess around with the values.

[robcat2075]

I was trying to use ever smaller values to try to exaggerate the SSS effect on the model little realising the opposite was needed.

 

I have done that.

 

I thought 1.0 might be "default"

[robcat2075]

making red largest lets more red scatter to the shadow side. Tricky to set the green and blue so the bright side doesn't look weird.

[robcat2075]

Here's a Steve Shelton model with SSS set at 4 3 2

 

[Rodney]

Nice looking SSS.

 

 

Not to run this off topic but...

 

Speaking of Steve... how's he doing?

I heard he was in pretty bad shape.

[robcat2075]

 

Speaking of Steve... how's he doing?

I heard he was in pretty bad shape.

 

I chatted with him briefly today. He is recovering, although I'm sure he wishes it were faster.

[Rodney]

I chatted with him briefly today. He is recovering, although I'm sure he wishes it were faster.

 

The next time you chat with him please let him know we are thinking of him, miss him and hope he has a speedy recovery.

[markarjun]

Here's a Steve Shelton model with SSS set at 4 3 2

 

StanSSS2.JPG

Those models examples are great and a lot closer to what I was aiming for, I would also wish your friend a speedy recovery. I'm guessing there isn't a way to have a decal drive the effect of the sub surface scattering similar to hair, as in paint areas of the ears and nose and leave other parts less effected so its less uniform?

[robcat2075]

... I'm guessing there isn't a way to have a decal drive the effect of the sub surface scattering similar to hair, as in paint areas of the ears and nose and leave other parts less effected so its less uniform?

 

We don't have that.

 

I have wondered if a compositing solution might work...

 

You'd render three passes. One is your "high SSS" settings. The second is your "low SSS" settings. The third is a grayscale map that depicts high vs. low areas and you use that to composite the two SSS renders into one final image.

 

It wouldn't be quite the same as the dedicated feature but might do well enough for character animation.

 

But it wouldn't be a lot more manual work since you have to paint a control map either way.

[Zaryin]

I cant find the original post, but I do have to post to the model here:

 

https://www.hash.com/forums/index.php?showtopic=46636&hl=alien&do=findComment&comment=399715

 

Hopefully the sss works about the same in v19 as v14. If not. Hopefully this will help a little.

[markarjun]

 

... I'm guessing there isn't a way to have a decal drive the effect of the sub surface scattering similar to hair, as in paint areas of the ears and nose and leave other parts less effected so its less uniform?

 

We don't have that.

 

I have wondered if a compositing solution might work...

 

You'd render three passes. One is your "high SSS" settings. The second is your "low SSS" settings. The third is a grayscale map that depicts high vs. low areas and you use that to composite the two SSS renders into one final image.

 

It wouldn't be quite the same as the dedicated feature but might do well enough for character animation.

 

But it wouldn't be a lot more manual work since you have to paint a control map either way.

 

Yeah compositing passes in post makes the most amount of sense.

Thats awesome many thanks!

I cant find the original post, but I do have to post to the model here:

 

https://www.hash.com/forums/index.php?showtopic=46636&hl=alien&do=findComment&comment=399715

 

Hopefully the sss works about the same in v19 as v14. If not. Hopefully this will help a little.

[John Bigboote]

Sorry guys- I was out on vacation last week. Here is my response to Mark on some Q's he asked:

 

Hi Mark- SORRY for delay responding- was out on vacation.

If it looks good in the model window then you are on the right track... you will need to make sure you have the SSS option activated for choreography camera/render. Try this: Go up to the top menu's (File, Edit, Project, View, TOOLS) Select Tools, Options (Cntrl-P) and then open the RENDERING tab... make sure 'advanced' is selected- then you can select where you will set your render options (kind of redundant, sure.) IF you select 'Use Settings From: This Dialogue', then you will want to root thru the settings to the left to make sure SSS is set to ON(10th option down) IF you select 'Use Settings From: The Camera', then you will need to hit OKAY and then open the render settings for your Chor's Camera and make sure SSS is activated there. I usually use the camera's settings, so I can change them quickly on the fly as I work.

Another thing to remember when in the chor... to see effects such as SSS, you will want to use the preview render button to the far-left (hover-over it should say "Render Mode 'Q'") and not the quick render preview mode 2nd from Left(Hover-over says "Render Lock Mode "Shift Q"')

Another thing to consider when using SSS is that the effect is highly reliant on certain lighting conditions, and that it prefers a strong backlight... also- experiment with the MP render and the other non MP as well)

I remember when SSS was a new feature, I was playing with it in a multipass render and I noticed that for every pass the SSS had to recompute- which made it costly in a high pass animation render. I asked the Hash programmers if a recompute was really necessary for every new pass, it seemed rather redundant- and they responded 'your right!' NOT needed! and today SSS is not as render expensive, a little hit, not bad.

I've noticed as well that sometimes if you have multiple SSS objects(or groups with different SSS settings) in the same scene they may not render correctly.... been meaning to report that.

Hope this helps and you will have something cool to show.

[markarjun]

Hi John many thanks for taking the time to give your insight into this. In the past I’ve found SSS quite hard to set up and to understand so I truly appreciate having all the help with this topic.

I haven't tried rendering multiple SSS groups, will let you know if i get similar problems.

 

 

I remember when SSS was a new feature, I was playing with it in a multipass render and I noticed that for every pass the SSS had to recompute- which made it costly in a high pass animation render. I asked the Hash programmers if a recompute was really necessary for every new pass, it seemed rather redundant- and they responded 'your right!' NOT needed! and today SSS is not as render expensive, a little hit, not bad.

I was amazed it was a lot quicker to render than i had imagined it to be! I think the render time was another reason for me avoiding it altogether.