problems with SSS in v17a?

[kikiriki]

I'm playing with SSS in v17a and I'm having problems getting decent results. I didn't use SSS in A:M before (but I did use it in other 3D software).

I found out that SSS doesn't work with "Global ambience colour". Can someone please confirm that?

 

I would be very grateful if someone could send me a complete PRJ file with working SSS material that looks like skin or wax (of course model could be a simple ball or cube...). This would help me a lot!

 

Maybe it would be good idea for one of the next contests, if the topic would be SSS materials. Now that A:M supports SSS materials, our community could use couple of good SSS materials.

 

Thank you in advance!

Dušan

[kikiriki]

(Duplicate Post?)

[Rodney]

A Sub Surface Scattering expert will be along shortly (Help... Nancy!!!)

I'm not sure enough about Global Ambiance Color enough to comment on that.

 

In the meantime, here is a post with project files for adding SSS to an eye. That is somewhat waxy.

 

http://www.hash.com/forums/index.php?showtopic=36034

 

If time is of the essense, here is the prime spots for discussions on SSS. (Sorry, I'm still trying to clean it up and organize)

 

Link: SSS Sub Surface Scattering (A search for all forum topics with the three letters SSS)

 

This topic has some good commentary on SSS and AO and some effective workthroughs/workarounds:

http://www.hash.com/forums/index.php?showtopic=41730

 

And perhaps most importantly, the Sub Scattering Forum is: HERE!

[Rodney]

Here is a comment by the initial programmer of SSS, Yves Poissant (Steffen added additional functionality and extended it into Materials):

 

There are many implementations of AO. Some implementations work somewhat like wikipedia describes it but not A:M.

 

In A:M, AO is a darkening trick only. From a shading point on a surface, AO shoots many rays in quasi-random directions over the surface and counts the percentage of rays that reach the background. If no rays reach the background, then the occlusion is 100% and if all rays reach the background, then the occlusion is 0%.

 

So rays that reach the background do not contribute any light. They just do not contribute occlusion. Nothing in AO contributes light but rather, wherever there is light, AO has the potential to darken it. The AO darkens whatever illumination was computed on a surface. So A point with occlusion of 100% will get black and a point with occlusion of 0% will keep its computed illumination.

 

Technically, AO is not a shadow either. Shadows are the result of specific light not reaching some area of the scene. Shadows do not darken some areas of the scene. Whatever light was already computed from some light stays there when shadowed from another light. While AO does darkens part of the scene.

 

AO will likely not work very well with SSS because they tend to compete one against the other. SSS tries to diffuse light in small parts of the geometry but usually, those small geometries are also those which have the most occlusion. Typical example is an ear.

 

The take away that I get is that SSS needs real lights.

This would be is why Global Ambiance Color will not effect SSS (assuming it doesn't).

An immediate workaround would be compositing.

[John Bigboote]

Here's a test candle project: All in 17.0A 64

candle.mdl

candle.prj

[Rodney]

Very nice Matt!

 

 

 

I'll add this also from the Soulcage Department (although its a little off topic):

 

Faking SSS

[John Bigboote]

I've found it usually helps to add a light behind the object to 'coach' the SSS in cases where you really-really want to see the SSS effect.

[John Bigboote]

Now, I want to add an animated flame using Rob's distortion box idea...

[John Bigboote]

MERRY CHRISTMAS... early. (the included matcap was used in the candle holder's properties.) image was 9 passes with DOF 3:29 in V 17.0A/64

candle.prj

candle.mdl

[NancyGormezano]

Ooooo...purty

[John Bigboote]

For hoots I'm gonna let it go overnight and render 256 passes... here is after 9 passes. Probably won't look too much different! See how I get stuck on this stuff?

[Rodney]

A little melting at the top and I'd call that perfect.

 

Edit: I see you've already beat me to that!

[kikiriki]

Guys, you are the greatest! Thank you very much for all those informations. You gave me a lot of material to study, so I will probably need a couple of days before I could give you a feedback :-)

 

Looks like a lot has happening since I don't checking forum regularly. If I understand this "Fake AO" works as a separate pass and we need an A:M composite or After Effects (or something like that) to merge AO effect with image?

 

Thank you again!

[John Bigboote]

If I understand this "Fake AO" works as a separate pass and we need an A:M composite or After Effects (or something like that) to merge AO effect with image?

 

No, not at all! A:M does it at the very last moment- after the last pass. You COULD do it the way you mention, in a compositer by checking the box to save the fakeAO in it's own buffer(image sequence)

[NancyGormezano]

If I understand this "Fake AO" works as a separate pass and we need an A:M composite or After Effects (or something like that) to merge AO effect with image?

 

FakeAO is a posteffect created by Jenpy - you enable it when you render via camera/Output/buffers/Apply cameras post effect to render = ON.

 

It adds minimal to no additional time to the rendering time. Quite amazing.

 

You must also set the depth buffer = ON (important) when rendering

 

Works great in 16b PC. Currently there is problem with rendering with FakeaoCPU in ver 17 (32). It is reported to work in 64 bit (by a fluke). I don't think it is available for Macs. (not sure). There is also a fakeaoGPU version. I can't get it to work, but I believe others have (not sure). Depends on your graphics card, drivers, I believe.

Edited September 13, 2012 by NancyGormezano

[NancyGormezano]

Also, in case you need more data for SSS - I did some testing in this thread (probably done in 16b)

 

It becomes obvious there is no one set of values for SSS.

 

Values you choose will depend on the size of your model (real life size is best), which should have thick and thin surfaces (therefore sphere, cubes are not good for testing), as well as your lighting setup, and tolerance for longer rendering time.

[kikiriki]

Thank you guys! You are the best! It's a great feeling to know that I could always get help here in this forums.

[robcat2075]

Doing FakeAO in "post" in an A:M composite project lets you quickly experiment with settings in real time, which is good because it has many more settings to get right than regular AO does.

 

But it can also be used for final finished renders (in v16, at least)

[John Bigboote]

256 passes! (1 hour 13 min)

[Rodney]

Okay, now you've got me confused Matt.

 

At 9 passes the candle was melted just a little.

At 256 passes the candle is brand new and right out of the box.

 

Multipass is good... but...

 

Probably won't look too much different!

 

 

Is this one frame from a (melting) sequence?

[John Bigboote]

Yeah... weird, right?

 

I did a little flame, animated it with a distortion box... and am now rendering a sequence- frame 1. (5 passes)

 

OH- another trick... i made a gradient material for the flame... the top is yellow with a glow, the bottom redorange with 95% transparency. I have become a BIG fan of this gradient trick.

[John Bigboote]

Forgot- I rendered a sequence with flame!

candleAM.mov

[NancyGormezano]

Wonderful! very, very luvley! I like how the lighting flickers. Luuvley drips. SSS looks terrific.

 

(But, to be a luvley butt, it could use some blue core in the flame, and a more irregular, charred wick as camera gets closer. Might look better if the light behind wick didn't become so obvious as camera approaches. That's the problem with trying for realism...the work is never done)

[robcat2075]

A fine candle!

[John Bigboote]

yeah- there's a lot I didn't like about that one... also the way the wax at the bottom of the candle was way over-lit. I've got another render in the soup now. Photorealism IS a carrot on a stick!

[Rodney]

Nancy beat me to the commentary on adding a little blue to the flame. That would be nice.

 

Now you've got me wondering if negative lights (a white light or two or three with negative intensity) set around the candle would intensify the effect of the SSS.

[itsjustme]

Great candle, Matt!

[John Bigboote]

Now you've got me wondering if negative lights (a white light or two or three with negative intensity) set around the candle would intensify the effect of the SSS.

 

Ah! Interesting... I might just have to try that out...

[John Bigboote]

STILL not happy with this... my flame model is casting shadows, flame would not cast shadows- would it? I have blue set as the 2ndary color of the flame gradient but because it is 100% transparent... you don't see it too much. I also think the entire flame sits too high on the wick... oh- the wick now has a wonderful displacement material on it... good idea, Nancy!

 

I plan to generate this project when done to the Contributors Cue... when done, ha! This stuff is NEVER done! You don't ever really finish something in 3D... rather you run out of time or simply 'pull the plug' on it.

candleAM.mov

[Rodney]

That is amazing Matt. Very nice all around.

When you do unplug from this please post it to either A:M Films or the Show me! forum so others can stand in awe.

 

As far as the flame casting shadows you should be able to turn that off via Cast Shadows=Off... hmm.... but that is just per model.

I suppose for a quick fix you could have two copies of the same model in place and hide everything except the flame in one and only the flame in the other. Then set the flame to not cast shadows.

 

Perhaps you could make the main flame just more transparent? or mess with the falloff and ambiance blend in the surface properties? (That last one wouldn't do much for the shadow but would blend the flame in a little more at the source.

[Darkwing]

Ok that is seriously very cool!

[John Bigboote]

Thanks guys... Rodney- I do have the flame as a separate model- so it was not a problem to turn OFF shadows for it. Keeping everything as a separate model is helping for the light lists too. Tomorrows render should be final-vinyl.

[Rodney]

Awesome. Divide and conquer!

 

Looking forward to more awesomeness.

Did you end up testing any negative lights?

[Fuchur]

Really, really great looking candle!

 

See you

*Fuchur*

[John Bigboote]

Did you end up testing any negative lights?

 

Yes, I just gave it a try on the light behind the candle that was 'pushing' the SSS and it didn't appear to have any effect at all beyond zero. I think that shows that SSS is an 'additive' effect and can't be used inversely.

[John Bigboote]

To finalize this 'CANDLE' obsession with me, and make good on my threat... the project is now available in the Contributors Cue.

 

Here is the link to the final animation:

http://www.hash.com/forums/index.php?showt...t=0#entry376016

 

HERE is the link to the project in the CC.

http://www.hash.com/forums/index.php?showtopic=43698

[NancyGormezano]

That looks great! Excellent job on the candle, drips & wick model. The lighting & flame gradient works better as well.

 

As you said in other posts: you're done drinking...uh...dinking...with this project. Any more dinking & drinking is up to those who download the porject...uh project (and I haven't even started yet). Thanks for the project, Matt!

[John Bigboote]

Any more dinking & drinking is up to those who download the porject...uh project (and I haven't even started yet).

 

Yeah! And I would love to see other people's embellishments .