I try to use the volume shader with absorption.
My scene contain the volumetric sphere and the wall with refraction just (IOR=1/1). Camera look at sphere through wall.

On raytracing mode the result is okay: volume through glass wall is red.

On PBR mode the result is wrong: volume through glass wall is grayscale. This lose a color!

Please help to solve this strange trouble.

I can tell more. It is very important subject because this is a way to produce true attenuation (when refraction use and opacity=1) and multi scattering for any geometry. And may be SSS also.

My framework assume using two same object with same geometry. First object have the mantra surface shader, second object have volumetric shader with uniform volume property. So second object is volume with same form as first. It is imitation of internal volume.

Last step - solve this trouble. After solving we will get true caustics with colors of volume shader with respect for width of object. Using simple non-sophisticated shaders.

Yeah, something wonky is going on here. The only thing I got narrowed down was that the same happens with a minimal glass shader (nothing but one Refracted Light VOP), so at least it's not in the default Surface Model VOP jungle.

I'd suggest you send this one to support.

This is a limitation of PBR itself - secondary rays need to select a single sample from the ray march for shading, and in doing so they use the average of the opacity rgb channels. The same thing would occur with colored opacity for surfaces.

I can't say for sure when we'll have a solution to this, but you're right that it's unexpected.

Andrew

andrewc
This is a limitation of PBR itself - secondary rays need to select a single sample from the ray march for shading, and in doing so they use the average of the opacity rgb channels. The same thing would occur with colored opacity for surfaces.

I can't say for sure when we'll have a solution to this, but you're right that it's unexpected.

Andrew

Thank you for reply. I think this feature would be good way for make internal media of surface objects. Would be wonder if SESI will find a solution.