surfaceLuminance not supported

[vlado]

Hi, has anyone had any GI issues with the latest build in that it takes a long time to build and often crashes specifically when using the V-Ray attributes? Can anyone recommend V-Ray attribute settings to use with this node when rendering with GI?

Does this have anything to do with the surfaceLuminance node as such? Also, which is the latest build in question?

Best regards,
Vlado

[matteley]

Yes sorry I didn't mention it was specifically the surfaceLuminance node.

We've just tried it with the latest stable 20120401 release and now when using the vray attributes it no longer crashes, (as it did in the release from 5th Jan) however when using attributes other than 'disabled' during the GI calculation it takes 20 minutes+ to compute.

Can I send you the shader network we're using? (The file uploader is giving me and invalid file error)

[vlado]

You can email me the scene to vlado@chaosgroup.com - I have some suspicions about what may be causing it, but it would be good to check.

Best regards,
Vlado

[matteley]

Hi Vlado, did anything come to mind after looking at the shader as to what was causing the problem?

[vlado]

Yes, since the surfaceLuminance node does it's own ambient light gather, it multiplies with V-Ray's native GI and an explosion of GI rays happen. We probably need to disable calculation of the ambient light in the surfaceLuminance node for GI rays. For the moment you can also override the GI material for the object and use a material that does not include the surfaceLuminance shader.

[This was one of the complications I was talking about when I explained why I was reluctant to implement surfaceLuminance in the first place ]

Best regards,
Vlado

[matteley]

Thanks Vlado, I've tried your suggestion and it's fixes the excessive GI calc time however the raytracing is also pretty long, probably too long for scenes with several lights, using the default 8 samples and even when reducing to just 2. I can see now why you were reluctant to implement this!

Do you think it's likely there could be an alternative v-ray specific node available in the future as it's a pretty useful when creating more complex car paints?

[vlado]

Do you think it's likely there could be an alternative v-ray specific node available in the future as it's a pretty useful when creating more complex car paints?

Hm, I didn't realize you are using for car paint. Strictly speaking using surfaceLuminance not a physically accurate way to do a car paint shader. There are probably other approaches that could work better for physically based rendering, while still being artist-friendly... Can you explain a bit how you are using it and what effect you want to achieve with it?

Best regards,
Vlado

[matteley]

We're using it to create a two-tone car paint where the tone (driven by a ramp where we had connected our surfaceLuminance node) of the highlights is driven by luminance direction. So essentially where light hits the surface of the body panel you get a graduated splitting effect in the paint reflection colour.

[ivaylo.ivanov]

Your render times are big because you have too much subdivision for the reflection - 128, and low adaptiveness. In fact detaching the luminance texture and using a simple color do not make notable difference for me. The GI bug though, should be fixed in the next build. Try to test it when you can (without Vlado's work-around) and tell if it's now usable.

[seandunderdale]

using simply a colour in the reflection colour slot results in a metallic look, however it doesnt allow tinting of the blurry reflection based on luminance. Ive been having issues in general creating the look Im after using the surface luminance and the Vray dome...and Im not sure its the right way to go as its really slow to render even without GI when set to diffuse + direct in the Vray options.

We basically need a method of having the base layer in the carpaint tinting based on the highlights from the HDRI. So we'd get a coloured specular effect in the blurry reflection that we can map with a ramp, and control the falloff. So, where the HDRI reflection is super white, we can map with white, then yellow, then red for eg. But driven by the HDRI reflection intensity.

Ill post up any results we get if we find a solution.