[rivoli]

what i'm wondering is, will it know how to bend light correctly when passing through inverted normals?

i mean this is how an inverted normal looks to maxwell (the black faces in the cube on the right):

[kingpin]

Thomas An., thanks for nice pics.

You are absolutely right about the path of refracted ray. However, the problem is that maxwell does not bend a ray if ray hits poly from behind.



The path should be one shwon as red line, but in maxwell, the path of ray seems to be interporated as yellow line (missing two refraction since hitting poly from behind).

I guess that Mihai's original test (No3) is the closest reality, but still off, and this really should be corrected especially since maxwell is sold as an optically correct renderer, and if NL corrects this, tom's model should also work.

Kingpin

[Thomas An.]

I think we are done This is an impass.
None of the solutions thus far are absolutely correct (they come close, they approximate, but they are not 100% bullet proof)

We need divine intervention from Oscar programatically
Maxwell needs an internal change.

OSCAR, can you comment on this ?

--

[tom]

I guess that Mihai's original test (No3) is the closest reality, but still off, and this really should be corrected especially since maxwell is sold as an optically correct renderer, and if NL corrects this, tom's model should also work.
Kingpin

Oscar, master!
I hope you're reading this IOR challenge.
We also know other render engines are the same....
Unfortunately we need few more lines of code in the engine.
I hope you have an idea on this important situation.

At least I can say that,
"If this can be done somehow, nobody will be able to compare it to other engines anymore!"

[DELETED]

DELETED

[kingpin]

After observing few glass filled with scotch and drinking them, I started to think that the problem is much more complex than I am thinking. Yes, maxwell has a problem handling a ray hitting from behind... This really needs to be fixed, but how? If maxwell is a regular raytracer, it can be fixed by adding feature to refract light hitting from behind... but this is not your regular ray tracing renderer...

One thing for sure is that NL has to come up with genius way to handle this (I am confident that they will...)...

Mmm.. I guess their love for water will hit new low.

Cheers,
Kingpin

[Aldaryn]

Now I'm really confused. Look at theese pictures: Almost like maxwell renderings, both are photoreal, and both act strange, where the glass and liquid meets. You actually can see the thickness of the glass...



in the first picture, you can see an oil candle, now, isnt that starnge? Fist, I thought, that its becouse of the retangular shape of the glass, but just look at the next reference, vine in a bottle, and at the neck, you can clearly see the thickness of the glass, but I dont think there is a gap between... )

So, Ive got this only possible conclusion: Maxwell became so photo realistic, that now, realitiy itself is adapting to it...

A.

[rivoli]

i bet oil's ior is not 1,33, rather something around 1,5. is definetely a faster medium than glass, but not as fast as water, so light crossing the bonduary between glass and oil would bend differenty than when it goes from glass to water. so refractions should look different as well.

[tom]

because it's rectangular and so thick and it's not about IOR rivoli...
try put some coke or whiskey or beer inside.. it won't be different aldaryn...
look left bottom corner of the curvature where the oil meets edge
it's the same about the wine bottle...
i think it's all not about IOR and mostly affected by circular shape...
rectangular glasses show their thicknesses easily...

[rivoli]

because it's rectangular and so thick and it's not about IOR rivoli...

well, just looked for a couple more images online, it seems you're right tom. my bad. can't compete with you when glasses and stuff come into play.

[Aldaryn]

...but the wine bottle is circular, so, I really dont get it... maybe its also related to the thickness of the liquid inside? Ill try to repruduce this effect with different glasses, and liquids...

Btw, tom, youre right, in the first one, the rectangularity is the key, Ive tried to fill it up with some wine, and got the same result. Wine tasted a bit odd after.

[bader]

I am trying to do like what (tom) and (Mihai Iliuta) glass and water . u know trying to catch up.

That’s what I did for both glass :
Alcohol: 0.35/ 145/1.36
Glass:0.25/151.152
i know it is bad

[Thomas An.]

... but just look at the next reference, vine in a bottle, and at the neck, you can clearly see the thickness of the glass, but I dont think there is a gap between... )

Interesting thing is that in your wine bottle you see thickness in the neck area only but not in the body. I suspect that thickness + radius of curvature have something to do with it as the phenomenon possibly relates to complete internal reflection (fiber-optic effect).

In the photos bellow the glass with wine shows no thickness, but the glass filled with opeque material does.



As you can see, this glass is pretty thick near the bottom and yet the thickness was not obvious when it was filled with wine...

[tom]

I am trying to do like what (tom) and (Mihai Iliuta) glass and water . u know trying to catch up.
That’s what I did for both glass :
Alcohol: 0.35/ 145/1.36
Glass:0.25/151.152
i know it is bad

it seems you have wrong normals of glasses bader.... try inverting them....

[tom]

thomas, keep it up! we'll find how in the end...hehe