Light select with SSS material does not match lighting pass

[sharktacos]

Yes, that would be a good upgrade when the main issue is fixed. We will keep you posted.

Maybe I'm crazy, but is it possible that it has been fixed?

I did a render just now and it looks like the lightSelect set to "diffuse" is only returning the diffuse contribution of a SSS2 material (as it should) and leaving the SSS out.

[tashko.zashev]

Maybe I'm crazy, but is it possible that it has been fixed?

I did a render just now and it looks like the lightSelect set to "diffuse" is only returning the diffuse contribution of a SSS2 material (as it should) and leaving the SSS out.

There aren't any changes made to this .
Hmm, I think the issue is in the difference in comp between the Light Select Diffuse/s and Lighting element when SSS2 is present in the scene. Can you try to render that scene with older build?

[sharktacos]

I got this on the max side, the fix was to render the light select in diffuse mode and use the diffuse filter to divide it to make a new raw - you'll get the sss bits in there. Admittedly it's not perfect since the result is sss + whatever diffuse was also in the material but it might be good enough for you in this case.

So I did a test of this, and it is not working for me. I made a sphere with SSS2 on it, put one light on it, and set up a light select for it. Then in Nuke I did
lightSelectRaw / diffuse, expecting this to look like the SSS pass. Instead it was blue. Note that on the SSS2 material the diffuse amount is set to zero.

Am I doing something wrong?

[joconnell]

Righty - so light select diffuse / diffuse filter gives you the raw light hitting the surface of your sss object - I used the ketchup preset so my object is red. My resulting light is blue since the red bits are absorbed and scattered, the blue bits are bounced back. My sss2 divided by the diffuse filter gives me the raw sss.

The light select raw only includes the direct hitting diffuse light, it won't contain any of the sss light scattered in your object, you'll have to also get some of the data from your sss2 pass. Here's a wee recombine and grade from vray 3.2 max - https://www.dropbox.com/s/b2e0jlpats...sss_2.zip?dl=0

[sharktacos]

Thanks for the file. Here's where I'm still stuck. what we want to be able to do is use lightSelects to say: "Here are all the passes from the key light: lighting + sss + spec + reflection = key light beauty. Now here's the same for the bounce light: lighting + sss + spec + reflection = key light beauty. Now we add those two together at the end"

The problem is that the sss pass would have both the key and the bounce light in it. The lightSelect just has one, but it is missing the SSS contribution. Your example only has one light, so I can't see how this would function when there was more than one light (hence the need for a light select).

Knowing you, you're a super smart dude, so I'm guess that your solution works, and I'm just not getting it yet. So I'm still prodding

[jasonhuang1115]

From what I can remember, the light select RE has been like this from maybe 2011 when I asked if it can carry the SSS and/or GI info on per-light basis....Guess we will get there one day....

[joconnell]

Yep I getcha shark - As Jason's said unfortunately some of the passes don't separate out yet. Gi is stored as a single total result at the minute and it's the same for the sss passes - a lot of it is a throw back to the GI method used as the irradiance map doesn't calculate individual direct light contributions, just a total amount of energy fed in to start off the bouncing. It's the same with sss too. What's changed though is all of the speed ups in brute force. Brute force GI and also raytraced sss has to take individual ray samples which do trace paths to a single light source so it's far easier to remember where each bit of light sampling came from.

The chaos guys are aware that people want it and are likely looking at it in the background, I'd say it'd be a bit of an overhaul for some of the passes and materials too so it's a fair bit of work but they'll get there soon enough and we should be able to do something quite similar to maxwell's multilight.

[sharktacos]

The approach we use to get around this limitation in Maya is to use render layers. You just put the lights you want into a render layer and toggle on the regular lighting, spec, and sss passes. Only drawback (which is why I'm hoping for a fix to the lightSelect RE) is that it increases render time since it is a render layer (thus meaning it is rendering the image again from scratch) as opposed to a render element (derived from the render).

[joconnell]

Yep, it's a pain in the ass to manager all those renders though - I'd say the guys have it on their to do list.