[rivoli]

However, I can tell with certainty that the refraction is not correct. As I see in the (upcoming) real-life photo the rectangular block on the right is actually seen on the left side through the physical glass+liquid setup.

do you mean something like this?

[Mihai]

Mental Ray?

[rivoli]

yep. and as long as i can tell that is the correct refraction we should get from a liquid into glass. you can easily test it yourself with a pack of cigarettes and a bottle of water (which is what i just did).

the one i posted above has a Hybrid IOR (1,17 for the inner part of the glass), while this one has 1,33 as IOR, but it doesn't look correct to me:

[Mihai]

you can easily test it yourself with a pack of cigarettes

What brand?

Are you using the l_glass shaders? Or just simply cut out the glass where the liquid meets and using phong? I'm going to test this with the l_glass shaders to see....

[rivoli]

gauloises here, and it works just fine (blue ones).

i'm using the physic_phen (it should be the same) and i cut the glass away where glass and liquid meet.

[rivoli]

here's another one. this time glass is not cut away where it meets water.
the glass is one closed mesh and the liquid another closed one perfectly coplanar where the two meet.
glass IOR 1,56
water IOR 1,33

mental ray glass lume shader:

[Kabe]

Yes, which makes me wonder why there are no reflections (in reality) between the glass and liquid? If someone could explain that to me please

There *are* reflections between glass and liquid, they are just not so obvious - but you can see them, if you use lead chrystal glasses for instance.

Here's % reflection at the different angles between a dieelectric of ND 1.4 - ie glas - and air according to the Fresnel formulas:

The light green curve is the amount of reflection of the ray entering the glass, the dark green ray is the leaving ray.
You see two things here: At 90° angle each dielectric reflects 100%, no matter what direction the ray travels. At 0°, there is already some reflection. And, starting from 46°, a ray can not leve the dense medium and get's totally reflected back into the glass.

So here's the situation between a liquid of ND 1.3 and a glas of ND 1.4:

So, as you see, you still have TIR, starting at 68°, and there's nearly no reflection for rays that are entering the glas from the liquid.

hope it helps

Kabe

[Mihai]

Excellent, thank you Kabe

[Thomas An.]

Hi everyone.

I have uploaded the experiment scene here:

(2.13Mb)
http://mywebpages.comcast.net/ThomasAna ... d_Glass.7z
This is a 7z file; as it has the highest compression and I need to be frugal about server space. You can open the 7z file using the Power Archiver utility.

The archive contains:
1. The scene as a 3ds file
2. The scene as a 3dm file
3. Two photos of the actual setup
4. Brief instructions.

I will upload the results here when they are done (I haven't started doing the actual comparison tests yet). If anyone else is interested you can try too.

--------------------------
EDIT: A few more 7z utilities

• http://www.7-zip.org/ (freeware)
• http://www.zipzag.com/features.html

[/color]

[rivoli]

wow, that's really great thomas. thanks for all the efforts you're doing to make this glass/water stuff clear.

i just downloaded and unzipped the file, it all looks very professional. let's start having some more liquid 'n' glass fun.

[Thomas An.]

These are some of the major previous discussions on dielectrics.
(I put them here so that they are cross-linked and easier to find)

http://www.maxwellrender.com/forum/view ... ctric#7858
http://www.maxwellrender.com/forum/view ... sc&start=0
http://www.maxwellrender.com/forum/view ... sc&start=0
http://www.maxwellrender.com/forum/view ... sc&start=0

[Thomas An.]

I still wonder about one thing regarding your method Thomas. I recall tom did some tests a while ago, with I think it was a dielectric inside a dielectric, like in your method, and the one inside didn't have any reflections at all.....I think that's what the problem was.....maybe tom can recap what it was about.

Yes, I remember that discussion... but... without a reference object I prefer to be sceptical about labeling something as "false".

When it comes to physics there have been a few instances in the past that my perception of what should be happening was not synchronised with how nature really works. (It happens a little more often during physics courses... we get stumped by the unexpected).

[rivoli]

well, i just found out the "hybrid" method does not work at all. refractions are all messed up and doesn't look as they should. it's good to know it.
thanks again thomas.

edit:
from what i've tested so far (hybrid's, mihai's, thomas' methods) thomas' glass of water is the one that matches closely the real thing, both in terms of reflections/refractions and caustics.

[Thomas An.]

Preliminary results. It looks like the latest method developed in the beginning of this thread is quite accurate.

[Thomas An.]

Rivoli,

Thank you for taking the time testing this scene from your end

Also, your tests with Mental Ray are interesting too. So Mental Ray gives different results by using the hybrid-ND method. This basically gives further evidence that Maxwell engine's inner working are fundamentally different than other renderers.