Maxwell Render | Official public forum

Afternoon all!

I'm facing a bit of a material challenge here. Has anyone ever attempted to render dichroic coloured glass (to be used
as facade design elements in form of external horizontal and vertical fins) The material changes the transparent colour
as well as the reflected colour depending on the view angle. The transparency goes from pale blue (oblique angle) to
saturated green (perpendicular) and back to pale blue. The reflected colour does the same from gold (oblique angle)
via a strong purple (perpendicular) and back to gold again.

So what's the best approach here? Coatings, fresnel, dispersion? I also hope to find a solution which doesn't need a
week to clear up or to reach SL 17. It needs to have a proper refractive index as it needs to hold up in close-up
renders. See the attached link for a more detailed product description.

http://www.schott.com/architecture/engl ... ng=english



Any input will be greatly appreciated, many thanks in advance guys.

Jost

coatings with an appropriate Nd should do the trick... check this out:

http://maxwellrender.com/forum/viewtopi ... 93#p291493

Best,
Jason.

Hey Half-Life,

The image for the link you posted is not showing correctly.
Is anyone else having this problem?

soarchitect wrote:Hey Half-Life,

The image for the link you posted is not showing correctly.
Is anyone else having this problem?

Yep, same here. Also a shame that there hasn't been more feedback so far, is everyone hibernating already?

Hmm, it was up yesterday for me -- it may be the webhost for it is having technical difficulties.

The basic idea was to show the color the coating will give at different thicknesses -- it uses a concentric-type gradient to map the thickness of the coating... here's a very primitive reproduction:

http://www.spotoarts.com/jason/coating% ... 0scene.zip

The gradient is going from a thickness of 100 to 1000 in a clockwise rotation.

The idea is to find the thickness of the coating to get the coloration you need you just find the degrees of the gradient (that matches the color you want) and multiply by 0.7111111, which will give you the RGB value of that spot in the gradient map. Then multiply that number by 3.515625 and add 100 to the result to find the thickness desired for that coating. for example:

If you want the color located at 45 degrees in the gradient then multiply 45 x 0.7111111 which gives 31.99999 (we can round to 32).

Take 32 and multiply by 3.515625 which gives 112.5 (we can round to 113) and add 100 since that is the smallest value in the thickness gradient and we end up with a desired thickness of 213.

You can change the values in the scene to narrow it down to a specific range of hues if you like.

Best,
Jason.

hey 1/2 life

Can you explain a little more your equation ?

thks!

well first I guess we need to establish a few facts, which may already be obvious but they are the variables we have to account for:

1) there is 360 degrees in the concentric gradient map I am using(which in Photoshop is known as an "Angle Gradient" for obvious reasons).

2) there are 256 shades of grey in the gradient (which is standard for single channel 8bit images, ie greyscale).

3) I have the thickness parameter set to go between 100 and 1000 -- for a total of 900 possible.

Since it is visually pretty easy to navigate a circle based on degrees we choose this value as our main means of determining what we are after.

So, the first thing we need to know is how to measure each greyscale value in relationship to our desired angle. To do this take 256 and divide by 360 which returns the number 0.7111111, which represents how far we are moving in "greyscale space" per degree.

Each value of grey in our 256 shades represents a particular coating thickness based on our material parameters -- since I set the parameters to encompass 900 possible thicknesses I will divide those 900 by the 256 and we end up with each grey value representing 3.515625nm of thickness in my material.

Because Maxwell does not allow a 0 thickness on coatings we have to account for the minimum thickness which in my settings is 100 -- since this value exists outside the extremes of the greyscale thickness map it needs to be adjusted into our minimum... 100 in this instance actually represents the value 0 in the thickness map.

I'm no math teacher -- so hopefully that is all clear.

Best,
Jason.

and all this would be so easy if NL implement plain "falloff" map.

for ND it was possibble to add some "manual control" with .r2 trick
but for colors? no . it is not possible.

clear now.
thks!

Everything is clear Half life except "31.99999 (we can round to 40)." Why not 32?

i also think, You can do the trick with coating, but i think that appearance of real material is in relationship with both - reflection and transparency of coating , and reflection and transparency of base material ... the best way to simulate real material is to simulate chemical composition with geometrical microstructure, but as told - it is not always fastest an practical way.

Oops, wow... big goof on my part -- sorry

As far as realistically simulating the surface -- the real surface is made up of multiple thin coatings on a glass substrate -- we can use an ior based coating to simluate the composition of one coating but we cannot stack coatings as we would in the real world so there is no way currently to accurately re-create the manufacturing process in Maxwell.

That said, I think the best approach is to try to approximate the visual appearance as best as we can given the tools we do have to work with.

Best,
Jason.

Yes, You are right Jason ... Maxwell is lightsimulator, not material nor chemical, but there is plenty of options for approximations.

Patrik

You know, now that I think of it it could be possible to get this effect absolutely accurate if you could model the glass and coating layers accurately but I'm not sure if any modelling package can handle accuracy down to the nanometer level... is that even possible?

What makes this work is the different index of refractions of the coatings and glass bending the light as it passes through each layer -- when you change your viewing angle the perceived thickness of each layer changes very slightly, bending the light waves slightly differently for each possible vantage point... but of course, unless you are doing an animation you would not be able to tell the difference between that and a good approximation.

There's actually quite alot of everyday modern materials that use these coating techniques. I only see that becoming more prominent as time goes by.

Best,
Jason.

Hi all,

Thanks very much for all the input and research so far. For a moment I thought my topic was being hijacked but now
I'm glad to see that it sort of came back. However, it seems that there's no easy solution to this rather tricky material
so I just pushed on and after a few tests I got something I'm happy with for now...




This is just the beginning of a complex facade study, it will get really exciting when I have to do the night time views
where those fins will be backlit with bands of LED lights, potentially introducing a fritted or printed layer between the
individual glass sheets. Should I be worried about my x-mas holidays...??

Again, thanks everyone and have a good weekend.

Jost

Hi all,

I have done a few more test renders and I'm more and more satisfied with the result even though the glass is
not 'physically correct'. This is my current setup for the dichroic glass using a gradient map in the thickness
channel of the coating layer:



...and here are some more test renders:




Now then, how do I do the best set-up for a LED based lighting solution? The idea is to 'backlight' the fins from
their edges. If you look in detail you will see that the fins (both, horizontal and vertical) are mounted into the
bronze framework. Is it best to use a simple single-sided polygon as a light source or maybe a long strip using
a mapped mxi with little white dots for each single LED position? Also, is there an emitter preset for LED's?

I'm just not sure what the best solution is for the current 2.1.0.3 version, haven't done this for a while so I'd really
appreciate your help. Thanks a million in advance.

Jost