Emitter plane and backlit pusbuttons

Sun Jul 15, 2012 8:51 pm

I have been experimenting with backlit pushbuttons, but have two issues that I cannot find a solution to. I am wondering if anyone has had the same problem? The first problem is that, as you can see in the image, light is not scattering into the button edges like it would in real life. It is practically opaque in those regions, and any settings for scattering have no effect on it. I can drop the scattering down to make those areas more transparent, but then I lose the scattering effect that I need.

The other problem is that making red pushbuttons with red emitter planes, they always come out yellow and orange, even if I take out ALL green and even add a bit of blue to drive the color over into the cooler side of the spectrum:

Mon Jul 16, 2012 8:32 am

This test seems to indicate that scattering does drop off too quickly. If the wall thickness is 1 unit thick, then the distance from inside corner to outside corner where any two sides meet is 1.4142 units. That's not even 1 1/2 times the wall thickness, yet the reduction in light that is able to escape the pushbutton at the corners is way too low. So clearly, at the top of the pushbutton where the "roof" overhangs the pushbutton body, light scattering doesn't have a chance in there, if scattering is diminishing as quickly as in the corners:

Setting index to zero helps a tiny bit, but not really any great improvement. It's almost as if scattering is only bi-directional, not like photons scattering in all 3D volumetric directions at once. If so, that would explain why the overhang is not getting any light scattered into it...when the photons are scattering, they are only scattering in two directions...back towards the source, and away from the source...not off to the sides?

Mon Jul 16, 2012 12:55 pm

Did you try to model slightly pyramidal emitters instead of simple planes ? You will get four normals casting some light in the angles of the button.

Mon Jul 16, 2012 5:29 pm

I'll try that out. It seems that might help, but there will still be an annoying opaque area at the "roof" edges, because tilting the normals will only change those ray vectors only so much. On the main body of the switch (the "box" area), I'm thinking there really should only be a slight reduction in light getting through at the corners, regardless of the shape of the emitter, in this situation.

I will post an image later today...Have an appointment this morning.

Mon Jul 16, 2012 9:58 pm

No change with the pyramid emitter. The pyramid sides are angled about 9.5 deg from horizontal, and the entire pyramid is rotated 45 deg so that the pyramid surfaces face the pushbutton corners:

Mon Jul 16, 2012 10:45 pm

Hi treddie,

Can you please post the scene of just the button? I would love to play with this a bit...
I have the feeling you may get better results if you reduce the scattering and increase roughness...

Cheers
Chris.

Tue Jul 17, 2012 2:25 am

Here you go. The first zip contains the ocs file, and the second, the obj/mtl (I couldn't get the site to take it all in one zip).

Incidentally, I had originally intended the emitter to be quite small, like a little bulb. The reason I went with a big square plane was due to trying to get this thing to work.
Which brings up the question, in specular materials, is roughness applied volumetrically or is it still just a surface quality?

Pushbutton 2.zip: (631.29 KiB) Downloaded 171 times

Tue Jul 17, 2012 3:34 am

I have the feeling you may get better results if you reduce the scattering and increase roughness...

You got me thinking now...I decided to dump scattering completely and go with absorption (but set to zero). I then went with very high roughness and set the index = 0. Then, in order to get a glossy reflective surface to the pushbutton, I mixed that material with a glossy blue surface. Emitter color is pure white.

This seems to work really well, but it still seems like a kludge to me. After all, scattering SHOULD work by itself, just like in a real translucent material.

Tue Jul 17, 2012 4:40 am

From a quick bit of research it looks like most of these are actually quite smooth plastic, with some kind of backing that does the scattering.
Therefore you get a nice smooth looking reflection off them, but they still have a nice smooth glow from behind.
I'm guessing this is what you were going for with the mix material - so could be worth a try? Add an additional backing to the inside of the switch with heavy scattering and roughness?

The best i could come up with in my short time was down the same road as you. Use absorption to help color the switch, but i did add scattering, with direction almost 1.0 (straight through).
This seems to come out quite nice - although perhaps not quite as "glowy" as you might like.

Cheers
Chris.

Tue Jul 17, 2012 10:06 am

That's actually really beautiful looking. I like it a lot.

On the material mix...The switches are smooth glossy translucent plexiglass with no backing, actually. So translucent that you cannot see through to any of the other surfaces. So they differ from most production switches I suppose. In my case, I was trying to copy that translucent plexi look. VERY very cloudy. Much more than in your example. Would you say you used a 50/50 split between roughness and scattering?

Basically, it appears from what you did and what I did that scattering has some severe limitations by itself. It still seems that the scattering algorithm does not really propogate well to the sides of any photon paths...In other words, it looks like scattering either goes forward or directly backward, and not straying too far from that 2-dimensional path. 30 degrees, maybe? I had the same problem in Maxwell, except that I gave up there since it took forever to see the results and make changes.

Based on what you did, I am going to play around with this more.