[Half Life]

Well, ideally you would want to use the Nd and attenuation of the transparent medium base -- according to what I found online silicone rubber has an Nd of 1.44.

This is a bit of text that I think reveals why I think it is essential to allow users to use Complex IOR for not only the medium of the SSS material but also for the "scattering" particles:

In general, light diffusing resin compositions are believed to have a higher light diffusion efficiency as the difference in refractive index between the binder resin and the microparticulate light diffusing agent becomes greater. With respect to the refractive index relative to sodium D line at 25° C., poly(meth)acrylate resins have a value of 1.49, polystyrene resins have a value of 1.59, and (meth)acrylate-styrene copolymers have a value of 1.50 to 1.55 whereas silicone rubbers before and after curing (uncured and cured compositions) have values of 1.40 to 1.45. Then a choice of a combination giving a greater difference of refractive index provides a light diffusing silicone rubber composition with a high light diffusion efficiency. In this regard, polystyrene resins are most preferred as the microparticulate light diffusing agent.

from here: http://www.freshpatents.com/-dt20110324 ... 069392.php

Attenuation does seem to be fairly low for this material based on what I found -- in the 1-2mm range... maybe more for special formulations.

Best,
Jason.

[tom]

OK well, the cure comes... Set your led emitter RGB 0,105,182 and turn on Color Multilight. Let's see how it comes, then I will explain why it works.

[Half Life]

You got my attention Tom!

I'm hoping my initial thought was going in the right direction and we are dealing with a Maxwell optimization thing.

Doh! I recognize that SSS material now -- that was from the workfiles of my Maxwell course, demonstrating how the various SSS settings interact and what SSS is capable of... I did a ton of those so I got a bit confused when I saw the settings and couldn't remember what I had done there and why

Best,
Jason.

[tom]

No, it's not about multiemitter optimization. The reason is the RGB<>Spectral conversion and sensitivity of purple/blue range in multiple scattering. Color Multilight -due to its nature- allocates more memory (and renders slower) for better color accuracy in these cases. Most of the time you won't need it but, especially for blue color it's essential.

[Half Life]

Outstanding information! This is something I would not have thought of, but makes perfect sense

I see several differently colored buttons on the controls here -- Could you start with any color and use color multilight or would it be best to always start with blue/purple and then switch to whatever color you needed (for best color accuracy and scattering)?

Also, am I correct in assuming this is because we are near the edge of the visible light spectrum here(next step being ultraviolet)?

Best,
Jason.

[tom]

Yes, it's very useful playing with Color Multilight for finding an initial emission color to be set. I mean I don't suggest hue shifting in the post.

[Half Life]

Gotcha, so if I understand correctly: I could set all my buttons up with whatever the "final color" I want and use color multilight to create a render that was more accurate for the specific wavelengths of each button (particularly blue/violets, I imagine this might be true for some reds too) but altering the colors or the multilight sliders during or after rendering would not be favorable for best results.

Since Maxwell uses 589nm wavelength for Nd I imagine a yellow/orange starting color would give the best flexibility for post render color changes... it might render fastest too.

This is exciting I've got a whole new set of tests to run now

@designpimp: BTW -- back in the day when I was doing lots of design work I thought about going by the handle "Pixel Pimp"... which I though was hilarious, so I heartily approve of your handle

Best,
Jason.

[David Solito]

Hello,

did you mean, Jason, making for exemple a render region of the emiter with color multilight on, see what's the best result and then launching the render with to good color value?
It's what I understand but

...but altering the colors or the multilight sliders during or after rendering would not be favorable for best results.

make me perplex...

[Half Life]

Well I certainly don't know all the math but that part that makes sense to me is that each color occupies a particular spot on the visible light spectrum -- what Tom shared is that color multilight forces Maxwell to spend extra cycles at each sample level calculating better light/scattering in the blue/purple range.

My extrapolation of this is that Maxwell is not as good at calculating colors near the fringe of the visible light spectrum (although I suspect given enough render time it will solve correct regardless)... Blue is very far on the spectrum from the baseline Maxwell wavelength of 589nm (which is how we derive Nd and K values) as well and my assumption would be that Maxwell calculates light that is colored at the 589nm range (yellow/orange) faster and better in regular mode than other colors as a opposite reaction.

I think this solution would be optimized to firstly calculate fully the lighting at the color of the starting value and most likely the farther you get way from the starting color in render or post the more it breaks down the accuracy (but again I suspect at high sample levels this is relative).

As I said I need to run many tests to verify all this and I cannot spare the time right now so it may be a bit before I post any conclusive evidence for or against.

To recap -- we're using color multilight here solely to optimize the render to focus more on (difficult to render) light color at early sample levels... to that end we would start and finish the render with the desired output color and ignore the multilight color swatch as if this wasn't using color multilight at all.

Best,
Jason.

[David Solito]

thanks

[designpimp]

No, it's not about multiemitter optimization. The reason is the RGB<>Spectral conversion and sensitivity of purple/blue range in multiple scattering. Color Multilight -due to its nature- allocates more memory (and renders slower) for better color accuracy in these cases. Most of the time you won't need it but, especially for blue color it's essential.

@Tom
AMAZING! YEARS OF WAITING... SO HAPPY! I should have posted a long time ago. Was this known or did you have to try?

I'm going to spend some more time on this tomorrow and try what you have suggested. I would have never thought of this! Could you send me this MXS Tom? I just want to use the same settings for sure.

@All
Regarding the attenuation question that came up... I think you are all right. I forgot that attenuation is supposed to be HALF of the light energy loss and not ALL. In general though we have samples of all different levels of opacities on these buttons in the office. In some cases I could imagine that 2-3CM of attenuation is definitely possible with the more translucent end of the spectrum.

I'm going to chop up some buttons tomorrow if I have time and shine a laser into them and see if I can measure how light attenuates towards the material we use in production.

@Jason,

Regarding the handle... totally not my choice actually! My friend and I purchased maxwell when the very first beta came out and since we used his email he created this account. I don't claim to be a design pimp at all.

BTW, if you guys want to see what these little SSS buttons do in real life (some swearing involved): http://www.youtube.com/watch?v=msdFDCcdwaA

or G-rated: http://www.youtube.com/watch?v=Y6bapWp-4k0

Enjoy!

[Half Life]

That was badass!

[tom]

AMAZING! YEARS OF WAITING... SO HAPPY! I should have posted a long time ago. Was this known or did you have to try? I'm going to spend some more time on this tomorrow and try what you have suggested. I would have never thought of this! Could you send me this MXS Tom? I just want to use the same settings for sure.

Yes, it's known color multilight processes more accurate color. But, it's a very rare case especially limited to blue emitters.
http://www.fileden.com/files/2009/5/30/2461193//led.mxs

[designpimp]

@Half life,

Thanks man! If you ever want some free gear, hit me up.

@Tom

Awesome... Tried rendering your MXS on my system. Good results. After I get back from a tradeshow this week I will try your settings on my scene.

BTW, I also noticed that when you move the multilight color to blue-red (purple) you get a gradient between more red in the color source and red on the borders of the SSS. Have you seen this as well?

Jim

[tom]

Glad to be helpful about it, thanks!

BTW, I also noticed that when you move the multilight color to blue-red (purple) you get a gradient between more red in the color source and red on the borders of the SSS. Have you seen this as well?

Yes, same here