[Polyxo]

Hi All,
I'm gathering all hints regarding rendertime effective setup of scenes lit by LED's. So far I've read everything I could find about LED-light on this forum. Also I did did a few tests based on the experiences of fellow users. However - my results are not yet there.
A displacement-mapping workflow is of course pretty straightforward and could be an option at times. But I will also need more accurate simulations and even some close-ups.
With modeled LED's I wasn't 100% sure that my setup is correct (unfortunately many of the images posted in older threads are no longer available). My main problem here was even and realistic looking light distribution. With a normal dielectric material for the "bulb" and a basic emitter the light appeared "too local". Would it be best to work with SSS materials for the plastic/glass? Or would it make more sense to use the shattering/diffraction options in the renderer to achieve realistic results?
Or does my model not match the requirements? Attached is schematic screenshot of my LED. Thanks in advance for any insights!

Holger

[Bubbaloo]

Cool. I don't have any answers or tips, but I am interested in your results. Can you show us some of your test renders?

[mtripoli]

I can't see the l.e.d. die in the image but make sure it is a simple triangle and, if I recall correctly make sure it isn't "touching" anything else otherwise you get into the emitter against an object thing... It looks like you have the whole inside of the led cored out. All you need is a solid with a small empty sphere in which the emitter is placed.

**EDIT**
In looking again the green plane is the emitter? That's not the way it is in the "real world"... stay with an accurate model and you'll get better results...

[mtripoli]

Here's a very (very quick - 3 minutes) render of an led... follow "real world" and it'll come out good...

[Polyxo]

[quote="mtripoli"]Here's a very (very quick - 3 minutes) render of an led... follow "real world" and it'll come out good...

Thanks mtripoli!
So how does your LED look internally?. I did mine pretty close to a schematic drawing I found online. Seems that they don't have completely identical inner structures. And how is its render-performance - did you let it cook for longer?

Here's some of my tests:


The rightmost row uses exactly the construction I posted first. The middle-row has no Anode and Cathode and also has its emitter oriented horizontally. The leftmost row has a triangular pyramid without bottom as emitter. Materials vary - I had best results with a single layer SSS-material (violet).

The second image shows single layer SSS materials with different settings. I also changed the geometry slighly again. The cavity now also has a domed shape (same offset distance everywhere) which lets the dark spot at the top of LED's disappear.

The best of those 6 LED's after rendering for a couple of hours. I'd still appreciate advise on improving the render-speed.

[mtripoli]

Here's a few images of the way I set this up. They're from some time ago so things may have changed (for the better) since then... The biggest thing, as I recall is to keep the emitter so it's not touching anything and keep it inside a "bubble" so all the indices work out right (someone please jump in on this point)...





After 1 hour:


With an Aperature map (just screwing around)...

Also, these are for a clear led body. I'm sure that to get a good diffused body would require SSS but I haven't needed these (so far)...


I'll post an image of two leds one laying down and the other standing up soon... (its rendering)...

[lebbeus]

I question whether the diffuse lighting effect is actually a function of SSS.
The base material is optically clear and I believe they only "roughen" the surface a bit to get the matte texture (if you cut one in half you'd still see the clear plastic).

using roughness instead of SSS should clear faster (while still producing the correct caustics), though I haven't run any tests of an LED material with 1.7 yet.

[mtripoli]

I question whether the diffuse lighting effect is actually a function of SSS.
The base material is optically clear and I believe they only "roughen" the surface a bit to get the matte texture (if you cut one in half you'd still see the clear plastic).

I beg to differ. I have never seen an led with a rough outside; in fact that would be very hard to maintain in production as an injection mold tool would polish itself over a very short period of time. Secondly, the led surface would be prone collecting dirt. Cutting a diffuse led would produce a shiny fractured face but if you look at the material under a microscope you would see the diffuse particles in the plastic.

I did a test using the "red sss" that is in the material editor under "Wizard - SSS converter". I used only the defaults. Here is the result:



More blue:


And a reference image I pulled off the net (not great):

[lebbeus]

found this site while searching today (free registration to download):

http://www.3dcontentcentral.com/Search.aspx?arg=led

don't know how the models are, though they are offered in a wide variety of CAD formats

[mtripoli]

Sunled has models in Pro/E and Solidworks... really nice models

http://www.sunledusa.com/

[mtripoli]

found this site while searching today (free registration to download):

http://www.3dcontentcentral.com/Search.aspx?arg=led

don't know how the models are, though they are offered in a wide variety of CAD formats

Search on "Tripoli"; I've posted a bunch of electronic IC packages on there in Solidworks format...

[KurtS]

some great photos

[Polyxo]

Sorry, it took a while, I'm too busy presently... Thanks All for great links and many thanks mtripoli for your test renders and illustrations! I changed my setup, used the form of a downloaded LED and used a simple reflector pretty close to mtripolis illustration. Things still need very long to clear up and I also get black dots for the very first time (1.7.1 used). There's two different material sfor this test:

• Front row: A simple, wizzard-created glass material settings default)
  Rear: The same material with a SSS layer applied

While it's possible to let the output appear kind of cool with the the post effects, noise in both cases is still a huge problem. All images are downsized (50%). Is there anything else I can do, also on the material side?

Thanks, Holger


After about one hour

After rendering overnight

Setup. I found no visual difference with/without the the Air material inside the reflector.