Nested dielectrics ... for dummies. Tweet

[EOA]

Hi, I am really confused by the 2020 manual and the explanation of the logic behind nested dielectrics.

It says: "Priorities are numerically assigned lowest (-100) to highest (100), where the lowest would be the object or material having the least desired effect on the image. In the case of the water glass example, the water would be assigned the lowest priority (1), then the glass (2), and, finally, the ice (3)."

I don't understand what "the least desired effect on the image" means. At all. To me there is no obvious logic to that order, or how to add further 'nestings'.

I get that the liquid and the glass should overlap each other - and I understand that, for some reason, in that scenario liquid gets the lower number. And I'm even less sure as to why ice gets the higher number though. And I am then totally lost when trying to then extrapolate from this and guess at where bubbles and condensation go on the list.

Let's take the following very real world setup I am trying to do: A glass (material a) containing a beverage (material b) with water condensation (material c) on the outside of the glass (material a), and bubbles (material d) in the beverage (material b) and also ice cubes (material e) with different bubbles (material f) frozen inside them. The last two (e&f) of which are partially sticking up over the edge of the beverage (b).

Can anyone please tell me the correct prioritization of all 6 materials: a, b, c, d, e & f? And/or explain the logic behind it.

[jayroth2020]

Sorry to see that section is confusing to you. I will take another pass at it to make it clearer for the next revision.

Think of nested dielectric priorities like "outside to inside," with the inside surfaces having the most negative priority value, and the outside surfaces having the most positive values.

So, in the glass example, the glass would have the highest priority, the ice the next highest and the liquid the lowest. In your condensation example, just shift things a bit, and make the droplets higher than the glass, and flow the changes through the rest.

In terms of bubbles within the ice, the bubbles would need a lower priority than the ice, while still being higher than the liquid.

It is helpful to allow for a range between each priority level to sneak things (such as the ice bubbles) into the mix. So, 100-80, 80-60, 60-40, and so on (or whatever range would typically work for you.)

[EOA]

...Think of nested dielectric priorities like "outside to inside," with the inside surfaces having the most negative priority value, and the outside surfaces having the most positive values.

Man, thanks for trying at least, I know it is tough to explain things. And I thought I understood it.

But, it is somehow profoundly confusing to me "what is inside what".

In the manual the ice has the highest number. And in your explanation here it has the middle one.

But that aside - I don't get why the cubes aren't actually the innermost thing and has the lowest number, aren't they mainly inside the liquid? (This is ignoring the whole, "isn't the cubes both inside and outside the liquid and the ice bubbles too"-part - which I might never understand)

I'd say in my (limited) understanding of "the physical world" the air bubbles in ice would be the innermost thing, then the ice cube along with the air bubbles of the soda, both of which are inside the liquid - and then the liquid is in the glass .. which has the condensation on it, which I guess at a stretch makes the glass 'inside' the condensation to my head. (but if you are stretching then you could also argue that the liquid is just a really big amount of condensation then.. like is the liquid even inside the glass - doesn't it sort of depend on where you are looking from? - the whole thing is severely ambiguous)

So in my head the priority would actually go like this:

0 ice bubbles
1 ice cubes
2 air bubbles
3 liquid
4 glass
5 condensation

but that obviously wrong according to both your explanations. I really need a diagram. Or a scene file. Or maybe my English, or my understanding of the world I inhabit is just severely worse than I thought.

[jayroth2020]

No worries. I don't have time to put a diagram together at the moment, but here is another version of the image used in the documentation, with added condensation on the outside of the glass. The condensation was accomplished via spheres cloned onto the surface of the glass object, as were the bubbles in the ice (hard to see in this image).

As the condensation is the furthest thing outside of the overall drink environment (consisting of condensation, glass, ice, ice bubbles and liquid), it has the highest priority. Next is the glass, which is followed by the ice, and, lastly, the liquid. The condensation has an IOR of 1.33, the glass 1.5, the water 1.33, the air bubbles just over 1.0 and the ice at 1.5 (to read a little stronger). The bubbles in the ice were given the same priority as the ice.

Hope this helps for now...

[EOA]

Thanks - it helps in the sense that I'll follow your advice. But I honestly still don't understand the logic behind any of it. To me the ice cubes being the outermost thing is only when I look top down maybe .. and so on and so forth. It makes little to no sens, except if the whole glass was frozen in a ice cube... to me and my language. I think there is a fundamental flaw in how its being explained, or a fundamental flaw in my brain.

Hope the bosses are investing millions in teaching and documentation as we speak.

[jayroth2020]

The priority method is less about the way things might be organized in the real world, and more about problem-solving a technical issue which does not present in the real world. So trying to figure out the why in this case by relying on real world experience is a distraction — it is NOT a real world problem, but an artificial one that occurs within the realm of rendering engines. The documentation presents the proper settings for success, based upon the way the mechanism of the solution works. Most engines offering nested dielectrics solve the challenge in the same manner (which a quick google search will reveal). I do agree that a diagram would be helpful. The next iteration of the documentation will contain such a diagram and offer more insight behind the Nested Dielectrics engine, supported by any useful points from this and other discussions.

[jayroth2020]

Here is the diagram. Each priority is called out, going outside to inside: The condensation is 4, the glass is 3, the ice cubes are 2 and the liquid is 1.

[AWOLism]

This is great! I have been somewhat confused too, so this more indepth explanation is very helpful.

/ Andreas

[bloomis2]

Respectfully reopening this subject because I'm constantly facing this situation with nested dielectrics, water and bubbles (working for cosmetics).
Here are a few renders. I think, according to your chart, that the correct one is the bottom left one, am I correct?

128 specular bounces, IOR for the water is 1.362 (shampoo) and 1.003 for the bubbles (air).
Would you reverse normals on the bubbles like I did on that second line?
On that same picture (bottom left), there are a few dark spots inside some bubbles, what do you think they are?

Thank you for your help!

[bloomis2]

In this new render with more bubbles it's easier to see the dark spots I was talking about when normals are reversed