[jfrancis]

I'm comparing a sphere of glass with a cube of glass carved into a sphere using nested dielectrics priorities and an "invisible" material. I'd love to get the two to look identical. Is there any chance I will be able to or is this as close as I'm likely to get?

[jfrancis]

Here's an image I found on the internet ^

My goal is to fill glass beads with interesting stuff and carve it all back to a bounding volume, like these valentine hearts above. I know I can make these beads in other ways, but I'm exploring the utility (or lack thereof) of this psuedo-boolean approach.

[Mihai]

I don't understand what you mean by "a cube carved into a sphere"? How is it carved into it when it's bigger than the sphere? I think you should get what you want with boolean operations and have the glass encompass the bits inside (and use priority 1 for the overall glass material).

Are some of those hearts made so that it's only the bottom that has a pattern, and the rest is glass? Some of them seem to be made that way....?

[jfrancis]

I don't understand what you mean by "a cube carved into a sphere"? How is it carved into it when it's bigger than the sphere? I think you should get what you want with boolean operations and have the glass encompass the bits inside (and use priority 1 for the overall glass material).

Are some of those hearts made so that it's only the bottom that has a pattern, and the rest is glass? Some of them seem to be made that way....?

I meant a cube carved down into the shape of a sphere by using a larger block of 'air material' surrounding an inner spherical cavity, so that that corners of the glass cube, which poke into the 'air material,' are rendered invisible.



I think they start out as a bundle of rods, each rod of a single color and through heat they make a 'millefiore cane' and then they slice those into millefiore beads.

[jfrancis]

I'm trying to solve what I see as the 'shared surface' problem by ignoring it and then carving everything back to a common shape.



And I'm interested in having the inner shapes 'reach' (or be cut back to) the surface everywhere, and not just be placed in a larger blob of glass like this paperweight.

[burnin]

One does not render an invisible material or even better, rendering time of an invisible material is 0.
Now we can go on bragging about we got the first volumetric RT engine capable of rendering physically accurate invisible materials.

Every material is made of matter. Even nested dielectrics (have at least different IOR)

You can still simply join the meshes and invert normals.

Or?

[eric nixon]

If you surround the subject by a small margin it will work.

[jfrancis]

If you surround the subject by a small margin it will work.

I tried that and was still seeing problem enough to make me not like the technique

[eric nixon]

Were you 'carving' into an sss material? that doesn't work yet afaik.

[jfrancis]

As a test, just the water example above. Maybe you mean even knocking the corners off a cube with a shower is too much and I should tidy up a shape with a thin shell very similar to the goal shape.

I'll try that.

[burnin]

yes, jfrancis is correct... there's something buggy going on
so i did a bit of testing, based on Nested Dielectrics V3 Documentation
and came to conclusion (please correct me if i'm wrong)
- nested dielectrics are unusable for replicating sort of complex stuff & materials mentioned in this thread
- on a bit denser geometry similar artifacts (as mentioned by OP) start to show all over neighboring/intersecting geo (visually looks like the terminator artifact) which makes it pretty much unusable for anything else but simple liquid in a glass, funny tho, no limitations are ever mentioned (in docs)

Well, i got some acceptable result with FIRE, but rendering it out of viewport become impossible, even Draft was off (either solid, translucent, transparent, SSS... nothing made any difference)


Done with V3.2.15

[Mihai]

Is this similar to what you're after?



It's just a boolean, but what will cause trouble, not for the refractions but for the overall look is the old caustics through glass problem. So I just made the bottom of the medallion a separate object, then what was left of the glass I set to be hidden from GI. This gets you the correct refractions but no caustics and so no darkening of the whole object (or the objects inside of the medallion).

[jfrancis]

I guess I'm looking for an interesting technique more than an end result. What you did looks pretty good.

You used a render boolean for that? I'm not sure I follow how you applied it. Or you mean you modeled it that way?

[Mihai]

It's just modeled with booleans. What's cool with nested dielectrics now is that it will work ok even if the red parts were made of glass themselves. That wouldn't have looked correct before.

[jfrancis]

Done with V3.2.15

What materials did you use in those images?