Compositing SSS buffer

[Seraph135]

I'm very familiar with the math involved to put Vray's buffers back together to recreate the beauty render...However, the SSS buffer seems to be a bit of a problem.

I'm using the fast SSS shader and I'm doing a test to try to figure out how to composite this buffer and recreate the beauty pass using all the other buffers. What I'm finding so far is this.

1. The SSS buffer seems to have direct lighting baked into it and possibly GI as well.
2. If the diffuse amount is set to 0 in the shader then all of its contribution shows in the SSS buffer. If there is some diffuse in the shader then part of its contribution shows in the diffuse pass as well.

I'm using version 1.5.12748_for_maya2009. The only way I've gotten the math to work is if I just minus the lighting, GI, reflection, refraction, spec (and anything else) from the beauty pass until I am left with what looks basically like an SSS pass, but a little different looking than what the actual SSS buffer looks like. So some information is being missed or included in one of the other buffers where it shouldn't be. Any information or a confirmation of this would be appreciated.

Thanks,

Tim J

[jasonhuang1115]

I am having the same issue. I sent a report to the Support alone with a Nuke script to demonstrate the issue.
The closest result I can get is, if there is only one light in the scene, using RE_Light_Select + GI + REFLECTION. But there is still a subtle discrepancy to the Beauty pass, visible when viewed in normal exposure level.
If I go the route of SSS + GI + REFLECTION + SPEC, then the difference to the Beauty pass is very obvious.
Here is a quick setup with one sphere assigned with FASTSSS2 and one spot light in the scene.
Comp tree:


difference of formula A


difference in formula B (5 exposure stops up)


I sort of remember that Vlado mentioned about this issue (the GI is calculated twice or some like that?) in another thread. Anyway, any insight from the dev. or other member would be appreciated.

cheers.

[R2J2]

using max so not sure how much of a difference there is but,

the SSS2 element does have GI in it
minus GI from SSS2, there still will be a small amount of difference so, minus all passes from the beauty/rgb to get leftovers and then minus that from the SSS2
I can get a match using this method but have problems grading elements before any of this math, so sometimes I don't worry about any leftovers

SSS2 is fast, looks great but I do wish the element was more like translucent refraction where, to my eye, it looks like only the translucent effect
SSS2 looks like it has diffuse and GI/lighting

[mr furious]

I believe this was a bug that was fixed in the nightly builds around 13076 (the current build we're using in production) and the SSS just "add/pluses" in. In earlier builds the GI was baked in...now it looks like the bty.

check out the change log in the release notes.

Andy

[jasonhuang1115]

I believe this was a bug that was fixed in the nightly builds around 13076 (the current build we're using in production) and the SSS just "add/pluses" in. In earlier builds the GI was baked in...now it looks like the bty.

check out the change log in the release notes.

Andy

Hey, Andy,
What Render Elements did you use to "add/plus" together? I am using build 13321, but I still got the same subtle difference between the composited result and the Beauty pass as shown in my previous post. Is any specific SSS2 setting needed to be activated? In my test, I simply create the material and choose a preset.

[mr furious]

Just used the standard "result" passes (no raw and filter). if your building it back up from the Raw and filter there maybe be subtle shadowing differences. I'll try and put together a simple test when i get back to work tomorrow.

A