[JDHill]

You don't need much space, just enough to ensure that faces are not coincident; it is important that the cavity have normals pointing inward, though. But, going by your description, I don't think I would try to embed the emitter at all. If you were making an LED, then yes, you need to do that, but according to what you originally wrote, I gather that the glass parts of your fixture are basically open-top glass tubes, something like a candle jar. I would just model things are they really are. If the glass itself appears to be luminous when backlit, it must either a) have so much surface detail (like a cut diamond) that light is reflected to the eye from a wide range of viewing angles, b) have a somewhat cloudy composition, or c) have a combination of the two. Regarding the second, here is a test, with a wizard-based "low grade" glass on the left, and a glass with slight scattering on the right:



The geometry is just a closed-bottom wavy (but smooth) tube, and the MXM has 2% roughness with no bump or displacement. Here it is: glass-cloudy.mxm. Not sure if this will be close to helpful...it's difficult to know exactly what you're after without seeing a reference.

[designer-made]

Been away from work for a few days and just read your post. Thanks for the response.

You're right. The reason the glass glows is because of irregularities in its composition. It is cast and intentionally has numerous bubbles of various sizes and what might be best described as internal "folds" or slumps, so it seems to send the light in every which direction. It is also not very clear so the SSS you set up seems a good choice.

I think the actual glass will be hard to replicate given its organic variation, but was hoping to get something reasonably close. Perhaps SSS with a little subtle displacement might do the trick. It's getting better little by little with the tips I've picked up here

Thanks again.

Cheers,