[ivox3]

Thanks Stu, ...

By the way, .....that's just the word I've been looking for, ....'collecting'.

.....didn't know what to call that property.

Well, ...it is collecting, ...but based on the reference photos --- it should also be evident in the middle areas.

....it's definitely the crux to this one.

C.

[ivox3]

(Just making sure)
Are you guys also modeling the internal cavity of the epoxy above the emitter element ?

I think I'll need your help on that one Thomas.

I only modeled it based on a physical reference. If there's something happening within the dome itself, ....then I don't know about it.

What is this internal cavity? ...as in a second surface for internal bounces?

btw: thx for the input.

[tom]

I think Thomas means you should model a cavity inside the LED first and embed the contents including the emitter inside it. Otherwise, penetrating the emitter directly into LED body would surely fail.

[Thomas An.]

I only modeled it based on a physical reference. If there's something happening within the dome itself, ....then I don't know about it.

Going by the illustration (which makes sense). Look at the red arrow ... it is pointing at a gray box. This box is a cavity. This cavity needs to be modeled as well (make sure the resulting epoxy dome+cavity is a single object and the cavity normals are facing inwards while the dome outer shell normals are pointing outwards).

[glebe digital]

I had assumed that the LED itself was directly 'potted' into a resin mould, thus no cavity Looking at the photos in the other thread this would seem to be the case........either way, I'll test this cavity approach so many thanks Thomas.

[dd_]

i read the illustration the other way around :/
ive now modelled the LED with a cavity like show i will do some tests

[tom]

I had assumed that the LED itself was directly 'potted' into a resin mould, thus no cavity

Cavity doesn't always need to be filled with an empty space. Something inside another thing with a reasonable tolarance can still be called a cavity because theoretically you can't make something touch another thing perfectly. This is the same thing we already discussed long about glass+water scenarios, too. However, if you're dealing with an emitter, you can't use it without least required faces of a cavity. Practically, it wouldn't affect the look too much. At least I suppose like that and even if this is not true, it's still technically correct than using an emitter directly inside an object without any reserved space.

[glebe digital]

Thanx for that info Tom.

dd, looking forward to the test.

[dd_]

here is the first test

screenie

you can see the cavity i used. i did try a bigger cavity but it didnt look all that good

[NicoR44]

I really like where this thread is going, so very helpful.
A great example to see that the more accurate you model the more accurate the result will be

Your renders are great Glebe digital!!

[dd_]

another test render

[lebbeus]

This is my first test, three different colors with RGB values taken from that LED paper that was posted here and cgarchitect a while ago. Emitter color is the same as plastic color. The two outer columns have sss (though roughness works well too). The three columns represent different lens shapes, created with parabolic curves (focus at emitter)…

after seeing Thomas An.'s post, will remodel tonight and try a spherical lens as well.

[dd_]

another test tweaked model, light catching more at front now and the middle beam of light is actually appearing from the middle of the caustic arc

oh and the LED is not sat on the floor

[dd_]

enclosed LEDs

[ivox3]

Cavity doesn't always need to be filled with an empty space. Something inside another thing with a reasonable tolarance can still be called a cavity because theoretically you can't make something touch another thing perfectly. This is the same thing we already discussed long about glass+water scenarios, too. However, if you're dealing with an emitter, you can't use it without least required faces of a cavity. Practically, it wouldn't affect the look too much. At least I suppose like that and even if this is not true, it's still technically correct than using an emitter directly inside an object without any reserved space.

Still not sure how to go about this, .... I'm not just going to model a cube, because I don't think it's accurate. Maybe an offset surface of the primary dome with an imperceptible air cushion between the two ....... ??

.....back to the lab.

___________________________________________________

Dave, ....excellent tests. I took another reference shot of an LED light pattern to start making comparisons.

Note: The addition of the reflector cone should be noted with this image. The other reference shots used a bare LED [no housing].



___________________________________________________



Lebbeus, ....also a nice test. Any chance of flipping those things around so we can see the light outpout(caustics) ?