[rivoli]

Do you mean "below 1"?

ops, sorry about that. yes, i meant below 1.

[Thomas An.]

Micha,

We know about the 0.847... but the method is incorrect; it was design for other render engines but not Maxwell.
http://www.maxwellrender.com/forum/view ... 272&#24272


(The explantion is all in the first page, ot two here; at the beginning of this thread)
http://www.maxwellrender.com/forum/view ... 06&start=0

[Mihai]

You only need to set the absolute IOR of the mediums (such as 1.51 for glass, 1.33 for water etc). You don't need to set the relative IOR of 0.85 for the glass>liquid interface since the raytracer should take care of this by itself . This is how the physics and l_glass shaders for mentalray work anyway.

[Micha]

Sorry Thomas, your method dosn't looks like Maxwell should look, for me it is a fake - two surface on the water surface - I'm not happy. I hope, this is only a temporary "best" way.
NextLimit should create some interface materials with two IORs, that would be easy physical correct.

[Thomas An.]

Sorry Thomas, your method dosn't looks like Maxwell should look, for me it is a fake - two surface on the water surface - I'm not happy. I hope, this is only a temporary "best" way.
NextLimit should create some interface materials with two IORs, that would be easy physical correct.

How is it better when you have to split the glass body in two and use 0.85 in the renderman engines ? Is splitting the glass something physically correct to you ?

dosn't looks like Maxwell should look, for me it is a fake

You are looking at the photo and then the Maxwell result and you are still telling me the result is a fake ?

I guess you really mean the method of construction...

[Hervé]

he means when he poured some water into his computer, it does'nt give a glass of water... very fake..

[Micha]

Thomas,

I mean physical correct like simulations in FEM stress analyses, each boundary surface between two materials get an own property, a simple rule. Splitting the glass - it is logical if you try to think in boundary surfaces and not in volumes. For me, it is difficult to understand why I should use two surfaces at the top of the water. Also, if I think in "volumes", it is not so easy to understand. This is not the wonderful simple usage of Maxwell.
I'm curious, what the NL team will do.

[Thomas An.]

For me, it is difficult to understand why I should use two surfaces at the top of the water

Micha,

There are no "surfaces" in my method. It is only solids. One solid for the glass and one solid for the liquid.

-

[Micha]

Do you not mean the schematic on the bottom of this page?

http://www.maxwellrender.com/forum/view ... c&start=15

[Thomas An.]

Do you not mean the schematic on the bottom of this page?

http://www.maxwellrender.com/forum/view ... c&start=15

Correct.
(The green surface you see there is only when you have colored glass, otherwise it is usually joined to the glass as a solid piece)

[Mihai]

That's the one, but what you see there are two solids, one inside the other. The liquid completely surrounded by glass.

Personally I don't know which method you can call more realistic because they each fake it, but in different ways. The point of this discussion is to show that currently only Thomas An's method gives correct results with Maxwell.

[Mihai]

Well here is my mentalray version. Looks very accurate as well, using the l_glass shader and glass using the "hybrid" method.

Rendertime 1 hour, setup time to get it to look a bit like the photo, around 3 hours.

It's nice to see that Maxwell has such accurate lighting distribution when comparing with the photo. The mentalray version is not as accurate, especially looking at the dark border between the box and the wall.

[Thomas An.]

Rendertime 1 hour, setup time to get it to look a bit like the photo, around 3 hours

Mihai,

Thank you for investing so much time on this.

The mental ray result is quite interesting, especially the fact that you accomplish all this from the hybrid-ND method (the inner engine workings are definately different)

The image looks very sharp (good contrast). Also, it is good we can actually see the bottom ridges through the base of the glass.

The caustic is not well developed; I can see it forming, but is being overwhelmed by the shadows of the straws which are a bit too strong.

You are right about the ambient lighting... it is somewhat off.

The image overall is good though and non of these points would be evident without a reference photo (we could be arguing for years about what looks right or not without the reference photo)

Again, thank you much for taking time to make this comparizon !

[Thomas An.]

Micha,

Maybe you can try this one using Rhinoman/AIR ? (if you have any free time these days ?)

That would be interesting to see as well.

[Micha]

Now I ask me, do I have overseen something - why is the hybrid method in maxwell wrong? In the real world, the boundary surface between two mediums is important. But why dosn't work this in Maxwell? Is the problem, that Maxwell dosn't accept ior smaller than 1? If yes, than is the hybrid method good and Maxwell need a small "bug" fix.