Maxwell Render

Maxwell uses the absolute Index of refraction at all times (which I think is more straight forward)

You can do the Hybrid-ND method in Maxwell too, but instaid ND=0.85 (or 1.11) you need to use 1.33.

Lets try it. I will try the Hybrid-ND method and use 1.33 for liquid boundary (I suspect the refraction will be correct.

_________________
"Only the happy can escape the labyrinth, but only those who escape are happy"

yes, it's correct. here's a "mihai" glass/liquid image. refractions looks correct (well, not completely, water top refractions are incorrect) but caustics are off.

Maybe we're getting confused again...to be clear, by hybrid, I ment part of the glass geometry (which in reality would be in contact with water) has a different l_glass shader applied to it which effectively makes that surface double sided, something not possible with Maxwell yet I believe.

So there are three different l_glass shaders: glass(1.55), glass+water interface (1.51 glass, 1.33 water), water(1.33). You have to set it up like this because that's how the shader works.

Tests like these show the strengths of Maxwell very well I think. You just had to set up the emitters in the same positions as in the photo, and you got correct light distribution, caustics and shadow transparency. In mentalray I had to rerender all the time while waiting for the photon and caustic calculation and without a photo reference you have no idea if the settings you use are correct or not.

It's true the straw shadow is too strong, which I guess comes from mentalrays implementation of area shadows. I suppose the stick/straw you used for the photo is completely opaque? That's why I didn't apply a transparent shadow shader to it in my pic.

_________________
Next Limit Team
Merchant of happiness

Right if you take that glass-liquid boundary and give it an ND of 1.33, instead 1.1 (which means we no longer use a hybrid-ND for that surface) then the refraction will show correct as Rivoli just proved.




It is plastic but very much opeque.

-

_________________
"Only the happy can escape the labyrinth, but only those who escape are happy"

Absolute ... I think the Maxwell glass use internal a relativ IOR to Air like the other engines. But, we will see. Maybe, you should try both, your absolute (?) and the classical biased renderer relativ ior (1/(1,55/1,33)). Good look.

_________________
Rhino3D and more - 3D visualisation for designers and architects ... my gallery ... Freelance. ... www.simulacrum.de

You are just being difficult. It doesn't really matter if it measures speed of light relative to AIR or relative to Vacuum. Lets call both of these cases "absolute".

ND air=1.000277
ND vacuum=1.000000

Maxwell uses the absolute ND all the time for every material and you do not need to define special "relative" IORs (such as 1/glass/liquid)



This has been tried over and over (we are going in circles). You have not being paying attention.

_________________
"Only the happy can escape the labyrinth, but only those who escape are happy"

Rivoli,

Thank you for the test. Do you have an explanation for the missing reflections on the top of the water? In which direction show the normals? Which IOR do you use for this surface?

_________________
Rhino3D and more - 3D visualisation for designers and architects ... my gallery ... Freelance. ... www.simulacrum.de

It's because in that pic, the ray enters the liquid directly, so when it exits the liquid at the bottom, it thinks it exited into air, instead of glass, because the normals of the bottom of the glass are pointing inwards. That was Thomas whole point of encompassing the liquid in glass, so the ray can keep track of when it exited the liquid and entered the glass.

_________________
Next Limit Team
Merchant of happiness

normals are pointing toward the camera, up that is (otherwise we'll just get a totally black surface). water top has 1,33 as IOR.
i do believe reflections are not the problem there (i know the image is not that easy to make out because of its size and low sampling level reached), as for the refracted stick, which looks almost correct to me. the problem are those refractions by the sides of the glass actually, those don't look correct at all. adding an upper surface with IOR 1,51 (as explained in thomas way of doing it) with a very small gap solve the problem.
actually i don't have any other explanation but the one already provided by thomas, which is based on snell's law, incident angles and the way light bends passing through different medium.
you better ask him directly though, he's the one who came up with it.

edit:
or the one mihai provided just above.

maybe my superb drawing will make it clearer why it's not correct in Maxwell:

_________________
Next Limit Team
Merchant of happiness

stunning drawing mihai. it looks better than the rendered glass actually.

thank you, thank you....perhaps i should change the green shade a bit....

_________________
Next Limit Team
Merchant of happiness

@Mihai: If you split the red surface in a water/glass surface and a air/water surface it should work. But if it dosn't work, than for me something in Maxwell is wrong, because than anything should be physical correct.

@Thomas: I only know my renderman shader internal code. Here, ND is 1 too. The calculation for the glass shader is internal 1/ior. So, if I want to use an other "from" medium than Vacuum (Air) than I must replace the "1" with the from_ior or I must use 1/(from_ior/to_ior).

You say, the Maxwell glass shader is allways absolute to Vacuum (Air), but what we need is a water shader that is relativ to Glass or a glas shader that is relativ to water. Maybe, we misunderstand us, because you are thinking in "volumens" and I in "surfaces". I would like to use boundary surface shaders.

OK, I give it up. In the moment I have no time to test it here too. Sorry. My english is to bad to talk about this.

_________________
Rhino3D and more - 3D visualisation for designers and architects ... my gallery ... Freelance. ... www.simulacrum.de

Exactly !
You are creating a hybrid IOR (specificaly for the glass/liquid transition). This is a hack to handle that situation. You do not need that in Maxwell.

Maxwel rays are more advanced and they are history aware (see first post). The deflections are calculated from the stack/memory of each ray and using the snell's law.

This does not mean that the Maxwell way is a bug. It is just doing the same thing in a different way.

_________________
"Only the happy can escape the labyrinth, but only those who escape are happy"

No. Maxwell is doing it in a different way.
Look at step 2 in the diagram:
http://www.maxwellrender.com/forum/view ... 06&start=0

_________________
"Only the happy can escape the labyrinth, but only those who escape are happy"