render pass problems

[sharktacos]

I'm trying to build together lighting passes in comp and am getting some ugly looking results from vray.

In the this example, I left out the spec (which also has errors which I will discuss in a later post) and am only looking at the diffuse result for now. To get this I multiplied the RawLight and Diffuse to get the diffuse lighting. The problem is the RawLight pass has clipped out values on it, which even if I clip them to 1 still gives incorrect results.

Here's the beauty. (Notice that the rim light spec on his head is very subtle here, but will get blown out on the passes):


and here are the diffuse and rawLight passes



Combined to gether with a multiply (and a clamp on the rawLighting) I get this obviously wrong result (notice the hard line in his cheek, and the shadow on his right shoulder):



I tried outputting a lighting pass instead, but this looks even worse:



What's going on?

I am working in linear space, with 16 bit exr renders, comping in Nuke.
I'm using Maya directional lights, and vraymaterial for the shirt and vray SSS2 for the head.

[vlado]

The rawLight pass gives you the amount of light that falls on a surface. For regular hard diffuse materials, the light bounces off the surface from the same point that it falls on. This is why multiplying the diffuse filter and the raw light works. For a SSS material, the light may exit from a different point, so using the raw light is not possible. Instead, you need to use the dedicated SSS2 element to get the component of the SSS material that is due to direct lighting. There is no "raw sss lighting" element for the moment, but you can get it by dividing the SSS element by the diffuse filter one.

Best regards,
Vlado

[sharktacos]

That would account for the character's head, but I am getting the same error on his shirt as well, which uses a regular vrayMaterial.
So the same behavior is happening regardless of whether I use SSS2 or VrayMaterials.

On both the diffuse is clipping, causing incorrect results in the comp. That clipping does dot happen in the beauty.

Additionally, the spec pass is significantly brighter than it is in the beauty.

It appears as if the light's intensity has much higher values in the passes than it does in the beauty.

[vlado]

Hm, do you have any color mapping applied to the RGB output? Physical camera exposure?

Best regards,
Vlado

[vlado]

Also, what kind of light are you using, and which V-Ray version?

Best regards,
Vlado

[sharktacos]

I have the color mapping set to Reinhard, however when I switch it to anything else (linear multiply, exponential, etc.) the result is unchanged. No camera exposure.

I am using Maya directional lights. When I switch the lights to VrayLightSpheres it is resolved. However, there is no vray directional light. The Vray Sun is directional, but not really an equivalent. Since it is made to be used with the physical camera, it needs to have very low values. In my scene an intensity of .001 is extremely bright. It also tints the light color to a yellow.

So I conclude the problem has something to do with how Maya directional lights are working in vray.

We have Vray for Maya 1.5 SP1

[vlado]

There was indeed an issue with directional lights, which is resolved in the nightly builds; if you would like to get access to these, please email me to vlado@chaosgroup.com

Best regards,
Vlado

[sharktacos]

thanks, will do.

[bernhard]

sss shader ?

When reading your post, I was curios if everything else works as well.... so I setup a simple scene and made a composition in ps as well as nuke. I didn't use maya lights, I rather stick to all vray lights - and so I tested this scene with one recVrayLight.

I couldn't find any error at the first look, but then I could see that the SSS Shader wasn't correct when adding all passes together.

snapshot:


Hmmm... not 100% sure and so I wanted to ask if that could be double checked.
It looks like that the SSS Pass includes already the GI pass.

So if I add the GI pass, I multiply two times and the SSS shader gets to bright.

Here you can download the 32bit images:
Download 32bit Images
Download PS Comping
Download Nuke Comping


Like I said, not 100% sure but would be cool to double check

thx a lot

[R2J2]

I use max but I assume elements are the same

you need minus GI from SSS2 to get the correct result
there will be a very slight difference so what i do is minus all passes from the beauty to get a leftovers pass then plus that result

the problem being that you can't then grade an element before all this, so i mostly don't worry about the leftovers. IMO it's not enough of a difference to worry about and prefer the control I get in nuke

[bernhard]

Hmm.. interesting... like I thought, it looks like the GI is already in the SSS pass. I just was thinking that the SSS pass shouldn't already include the GI pass ?!

Anyways... you "minus" the GI from the SSS - that's a nice idea. My first thought was to use a mask, but the minus-idea sounds much more convenient.
Thank you very much for your hint and how you deal with that. Highly appreciated !

[vlado]

In older builds, the GI was indeed included in the SSS pass. This is corrected now in the nightly builds.

Best regards,
Vlado

[R2J2]

oh cool, is this the same for max versions?
we haven't upgraded just yet

[vlado]

It should be fixed in V-Ray 2.0 for 3ds Max, yes.

Best regards,
Vlado

[bernhard]

Great... thx a lot Vlado !