Maxwell Render | Official public forum

Hi all,

I'm wondering if anyone has even been able to get more success in what I'm trying to achieve.

I'm trying to recreate a button we use at work quite a bit:


I know its a pretty big nono to place emitters behind SSS, but I'm not really sure how else to create realistic looking buttons for the ones I am modeling. I tried to make a SSS square and encapsulate a LED in a extruded body slightly larger than the SSS square. The extruded body was a 0,0,0 (completely black) surface to absorb some of the light bounces.

Here is an exploded view:


I think the silicone button itself looks good (or close enough) with the SSS:


However, as soon as I ever turn the LED on... its game over with noise on the button (purple pixels when emitter has no RED, etc). Even at SL 23 there is still some bad noise:

This looks much better than the 14-16SL level, when usually the noise is totally over the place. Most of the other materials (or non lit SSS buttons) are clear at this point.

I've been trying to do this, on and off for years with maxwell and was hoping with 2.5 I would get closer... however still looks like my approach is wrong or the engine just isn't set up for this particular use case yet.

Do any material experts have alternative ideas on how I can do this?

Thanks!

Hi designpimp,

Make the material the emitter instead of using a sphere (led) to light the button! That should do the trick
and it will reduce the noise too!

Do you model thickness for your button as in real life or it's just an empty shell?

@Teabag,

I've done this for LCD screens and backlit silkscreens and it works, however for anything that isn't a flat 2D surface the light looks too uniform which makes the button feel a bit fake. We also have dim controls under these buttons (in real world there are 2 LED's and TACT switch) which even exposes more of the SSS look. If I was to take this approach the only thing I could think of is using an image based MXI emitter were I would draw in the inconstancies (seems like a major hack though) and this would not work for off angle views of the buttons.

This particular product has a pretty uniform light distribution on the button but other products we make really expose a hot spot, see this drum machine:



@Tom,

It should have thickness.. I'll need to check again as I left this test at home. It does appear from the screen shot that the bottom is open. Although I just removed the support structure of the button using a boolean subtraction so I cant imagine that Rhino would have allowed this if it was an open polysurface. I'll confirm tonight but I'm pretty positive only the light block is a open surface... I only removed the support structure to reduce the geometry that has SSS. Usually the button looks like this:

A masked emitter should do the job much easier.

What do you mean by masked emitter?

Even if your button has a thickness I doubt the real thing is quite thin like that. Are you sure, you're accurately modeling the button?

@Tom,

I'm 1000% positive the button is accurate. The geometry is taken directly from the mechanical design thats used to create the tool for the "real thing".

tom wrote:Even if your button has a thickness I doubt the real thing is quite thin like that. Are you sure, you're accurately modeling the button?

Tom,

Just checked... the object is definitely a solid (see closed polysurface on right) and is 8mm x 16mm x 8.86mm. This does not seem super thin to me...



Would love to get your help on this... been using Maxwell since the beta and still have never been able to reproduce this button type... I try from every version. I can send you a MXS if it would help.

The SSS is from Half_life's library (THANKS!).



And here is the emitter setting:



and here is the light trap:



Any guidance?

Jim

This may sound stupid, but did you try matching the intensity of the "LED" emitter material to that of your scene lighting? The reason I ask is that is a very weak light and it may be that the priority issue is somewhat in play here... if you could up the intensity of the emitter material before render and then reduce back to proper levels in multilight then it may clear faster.

The SSS looks good without the back-light LED effect -- so I think it may have more to do with the way it's lit then the material itself.

Not having tired this myself this is pure conjecture on my part... but I am very interested.

Best,
Jason.

designpimp wrote:What do you mean by masked emitter?

If you don't want the light to be uniform over the button you can mask the layer that contains the emitter component the way you want it (two hot spots if the button has two leds and so).

It's true that it will look fake if the button is very thick though.

@Jason,

Was hoping to catch your eye!

I just did a test with the same geometry (but removed the top surface simulating the panel on the product). I duplicated the emitter and placed it above the SSS object and changed the color to grey and pointed it down. I also disengaged the image based lighting I was using before. Here it is a SL16:



The next things I'll try are:
- Removing minor fillet on top of button that mechanical guys have and make perfect 90 degree box. Maybe this is the source of all my problems?
- Checking mesh in studio and making sure that light trap box is not intersecting SSS geometry (shouldnt be though).
- Use triangle emitter instead of square (seem to remember something about this from beta days)
- Changing size of emitter (at moment I'm guessing a rough size of a surface mount LED)
- Using hollow shell and thinSSS

Any other hints?

Jim

Try lowering the att. distance from 6cm to 1cm. Just seems not enough attenuation for such a material, which can cause too much light bouncing around so more noisy.

I can say that if you want to try thin SSS you will want to use both a Scattering and Transmittance map -- for whatever reason thinSSS performs very nicely with images in those slots and very poorly without (ie plain material). I've done some back-lighting work for thinSSS here to give you a starting point: http://www.maxwellrender.com/forum/view ... 84#p337684

Attenuation does seem a bit high for such a thin material but my experience is that very thin real SSS works fine with with the exception of the dielectric caustic reflection issue: http://www.maxwellrender.com/forum/view ... 64#p340564

Still it could be improved -- this looks like it may be an older version of my SSS material which I updated for version 2.5:

1) I would probably start by setting the attenuation for the actual thickness of the button wall, which would mean the light would lose approximately half it's power before reaching us (which would be a good thing as I think we may be getting too much light)

2) You could then adjust your scattering power to control the translucency -- I would set it high to begin with and lower it until you reach a good balance.

3) I try to avoid the Asymmetry parameter until I'm nearly finished as it can be a quick fix that can mask other problems (I found this out the hard way)

That material was made to mimic the soft plastic of my wifes Jeep dashboard (I actually pulled a plug out and brought it up to the studio) -- It was also made to compensate for the limitations of V 2.0 SSS... so I kinda feel bad to see it still in use

If you want to PM me the MXS scene I'll take a whack at it in the morning and see If I can offer some concrete answers... but this does seem doable.

Best,
Jason.

yes like Mihai said the att. dist is wayyy too high.. I'd say 6 mm.. not 6 cm...