Roughness issue using nested dielectrics?

[miguel-rv]

Hey, I am playing with the latest add of nested dielectrics, and I wanted to know what correct way to dealing with different roughness values in the different materials.

Right now, just trying to recreate a similar glass with liquid scenario, when I enable Nested Dielectrics, and I give a higher priority to the glass, the roughness value of the liquid is ignored when seen through the glass directly, but not if there is air in between.

As you can see, the transmission roughness has no effect when in direct contact with the glass, however it works when there is air in betwqeen the two materials. Is this supposed to be like this?

[adrien.herubel]

When the water mesh intersects the glass mesh, and the glass mesh has higher priority, it creates an inactive boundary or 'false intersection'. (See the diagram in https://docs.arnoldrenderer.com/display/A5ARP/Nested+Dielectrics#NestedDielectrics-Arnold'smethod%E2...)

The only properties Arnold will take into account at this inactive boundary are the IOR and transmission scattering properties. This will cause the transmission roughness to be ignored on inactive boundaries.

To work around this it could depend on what kind of effect you are trying to achieve, it it is a scattering effect inside water, then I would try using transmission scatter instead. If it is frosted glass instead, then texturing transmission roughness on the glass itself would be more appropriate.

Transmission roughness is meant to model roughness at the surface for dielectrics, so its behavior in nested dielectrics is not well defined.