Idea: Adding IOR to Volume Shaders for Better Nested Dielectrics in Cycles Hi everyone, I’ve been thinking about how Cycles currently handles nested dielectrics—like glass filled with liquid—and I believe there’s a relatively simple way to improve realism and avoid common artifacts, without major e…

is this a similar aproach to this PR? https://projects.blender.org/blender/blender/pulls/118478

No it was a idea i had today.However then this closure shader system come to mind. I think as simple this idea is,its to difficult to implementing or maybe impossible due the closure system. Because everytime the pathtracer goes from one medium to another it could readout the IOR propertie into a …

If you want to hack it, I think this works well for the scenarios I’ve tested: max(IOR1, IOR2) / min(IOR1, IOR2) = interface IOR. In your case; 1.5 / 1.33 = 1.12 as the interface IOR. Very weak reflections and refraction almost the same, making things close to invisible. Matching the IOR will make…

[image] pixelgrip: The problem as always is the closure system that parameter changes inside the system The problem is mainly in the BVHTree. A ray would need to hit both surfaces at the same time, so that a correct ray may be thrown from some median position. Hitting just one surface than t…

Well, it was worth a try – but of course it couldn’t be that easy.

Still, if you don’t mind waiting for a render to finish… :sweat_smile: OSL will let you force this inside the shaders ( or at least, theoretically; as I haven’t tried this specifically). If you setup a script, that traces a ray up to a minimum epsilon distance, and get’s a filtered hit, then you c…

This sounds interesting.Since a OSL shader can be rendered at least with CPU i wonder if such “problem solver” can be used this way you described. I find still the idea quite simple and effective. What is needed for a correct Fresnel calculation.The incident IOR and the material IOR. If every mat…

[image] pixelgrip: no stacking and priority are needed You still need to deal with that, but inside the OSL… In the simplest case, you’ll have two different shader calls, and four trace hits, which you need to sort out, since you can’t pass any information between the shader calls. And that…

Realistically I think we can only wait until Lukas project @Joel_RVC posted gets ready. It’s a shame that Blender’s Cycles doesn’t support state-based shading like other render engines, where important context like the IOR stack can be passed to the shader. This makes things like nested dielectrics…

Adding IOR to Volume Shaders for Better Nested Dielectrics in Cycles

Support Lighting and Rendering

moony (moony) November 5, 2025, 6:00am 11

I experimented with nested dielectrics myself in the past and came up with this solution.

It only has limited applications and is very scene dependent - but it does kinda work.