[oz42]

Jost,

superb results, the multicoloured sheen on the glass edges is fantastic! What sort of map are you using for the coating thickness layer? Is is just a very subtle perlin noise?

Not sure what I'd suggest for the LEDs but tiny square polys should suffice? Merge them all together so they're one object and apply an emitter to the lot. There are a few LED materials on the mxmgallery (but I'm not sure any are 'standard') and there's a great post about modelled LEDs here http://www.maxwellrender.com/forum/view ... &start=120 but I don't think you'll be going to that level of detail!!

[Half Life]

I had to do a bit of searching but I have found some IES files for LED strip type lighting which may help for this project:

http://colorgloledlighting.com/applicat ... tion_id=11

This company provides installation guides and IES files on their download page:

http://colorgloledlighting.com/downloads.cfm

There are others out there so if you have a particular style of LED in mind you may have luck getting an IES file from the manufacturer.

Best,
Jason.

[Half Life]

I tried the LED IES files and set up an array to match the design of the "Monochromatic Strip Fixtures" ( http://colorgloledlighting.com/product. ... oduct_id=5 ) to test to see how it works in Maxwell for edge lit glass type effects and I think it was a success.

I used frosted plexiglass and only let it render for a little bit.



Best,
Jason.

[fuso]

Hi guys,

Thanks so much for your replies, comments and especially for the advise!

@ oz42: Yeah, I'm quite happy with it, too, at least for the daylight version. I was also going for the low poly/merged
mesh solution. Does one still have to multiply the wattage with the number of emitting meshes? I was thinking to merge
only the poly's for each fin to get a more defined and accurate result.
Had a look at that post and it's a shame that none of the renders shows up anymore. You were right, I cannot afford
that level of detail even though it's tempting. As for the gradient map used in the thickness channel, it is just a
standard RGB black (top) to white 800 x 600 JPG map. In 3Ds Max I'm using a planar UVW map but have scaled it quite a
bit larger than the object.

@ Half Life: Thanks for that, this is great! I might even suggest those to the design team to match my results!
When you said 'array', did you mean a row of polygons matching the LED distance roughly? Did you then merge them as
one mesh per LED fixture, applied an emitter including the IES file to each individual fixture? I'm asking because I have
not much experience with using IES in Maxwell, I guess I was always too scared ending up with weeks of render time.
In a previous design stage of this project I had rendered tests with more or less the same set-up, using a simple
rectangular polygon behind every single fin and applied a normal emitter without IES to it. It worked really well and I
ended up with the same result as you've shown in your post reply. The difference now is, that the design team doesn't
want to use frosted glass but those multi-colored coated fins instead. My concern is, that they won't pick up the light
especially considering that this is an office building which will be lit quite brightly in the evening. I said there'll be a lack
of contrast against that sort of backdrop, but who listens to the visualiser before seeing the proof.... tssss.

Cheers,

Jost

[Half Life]

I suppose the results will be slightly more realistic than a simple emitter plane, you'll have to decide what is worth the time.

The IES files for that particular fixture represent one individual LED and for that style fixture you can have 2 different types, narrow and broad focus. I have seen in the past people here trying to model an LED and if you are not seeing the lED itself it makes no sense to go through that trouble, the IES file will give the same characteristics as the LED.

To use an IES you need to assign the material to a sphere -- I would size the sphere to be about the same size as the LED itself... and then arrange a line of those IES spheres into the configuration matching the number and spacing of LED's for the fixture you want to simulate. This is where the technical spec's from the company come in handy. For added realism you may want to do a simple model of the tube the would sit in, but it isn't really necessary. Then I would just install them as you would in the real world... just make sure you account for the spacing between the LED's and the edge of the glass as it would be in the real world. Frosting the edges of the glass may or may not help the effect (personally I would add roughness and anisotropy to the edges for realism)

The only real tricky thing here is you cannot use instances with emitters so with all those little spheres you will be generating a decent amount of extra geometry.

The end result will most likely not be tremendous with a "clear" glass... but it may add interest to the coating effect.

Best,
Jason.

[jwiede]

I suppose the results will be slightly more realistic than a simple emitter plane, you'll have to decide what is worth the time.
...
The only real tricky thing here is you cannot use instances with emitters so with all those little spheres you will be generating a decent amount of extra geometry.

After looking at the PDFs associated with those ColorGlo IES files, I came away with the impression that the IES was for the whole fixture, not the individual LEDs. The oblong shape of the IES domain in the PDF seems to confirms what was measured wasn't a single point source. Given an IES for a long, thin fixture (such as those in question), shouldn't the IES be applied to a long cylinder instead of a sphere? Should the ends of the cylinder be rounded or flat?

I've always been a bit confused about whether the emitter geometry needs to roughly approximate the actual shape of the fixture, or just be a sphere, cylinder, or whatever shape represent the appropriate "uniform distance" shape that contains the individual emitter elements of the fixture (in this case since it's a 1D array of emitters, a cylinder). I'd appreciate any help you could provide on that issue.

I'm guessing that since the IES includes the casings as well, there's no need to "encase" the emitter material when using such an IES -- is that correct? If not, should the casing represent the LED casing, fixture casing, or both?

(edit -- horrible light-related pun taken out back, shot, and buried in shallow grave)

[Half Life]

IES files are usually measured via placing the light fixture inside a sphere filled with sensors -- so a spherical object will best represent the environment the data was recorded in. The "real-world" shape of the fixture should be irrelevant.

The reason I think the particular IES files I used are for an individual LED is the "light model" that results and the fact that those fixtures can be customized to include many different numbers of LEDs. From that point of view it would make little sense to have one IES file that described numerous LED's that may or may not be what you are actually using... if that were the case then the IES file should be labeled with how many LED's are in that particular fixture.

Also in that fixture the casing (tube) of the fixture is optional so I do not see that as represented in the IES whereas the LED casing is part of the LED so I do see that as being represented in the IES light model... again looking at the light model in Maxwell Studio and the resulting material preview in MXED one can infer quite a bit about what is represented in the data.

This is based on the recommendations direct from the Maxwell manual and some web research -- I'm not an IES expert so if there is more information to be had I am all ears... but as far as I know the above is true.

Best,
Jason.

[jwiede]

if that were the case then the IES file should be labeled with how many LED's are in that particular fixture.

That's precisely what the associated PDFs do say: In the case of the Mono strip the PDF with the IES says 40 LEDs were used. I'm trying to find a utility to dump the IES file's data so that I can compare the numerical values with the PDF to confirm the data.

[Half Life]

yeah, looking at it now it does say that -- hmm... interesting. It still renders like a single LED not a long tube of LEDs... is that user error or a mistake in record keeping?

I'll be interested to see what you think after you get to confirm.

Best,
Jason.

[Half Life]

OK -- it was certainly user error on my part... that's what I get for not reading the IES file I read all the installation notes and skipped the IES notes, that's a big "duh" moment for me

Still it does make me wonder how are we supposed to use this file for fixtures with different numbers of LED's?

The IES file does render in a circular pattern that does not suggest a tube at all and the scale definitely says individual LED... I do wonder if there is a problem somewhere with the file itself.

Best,
Jason.

[Half Life]

Here's a render of the IES file output -- notice the white hotspot on the center... this was the narrow focus IES (6000K_30°_ITL56284) which should be more cylindrical if this is indeed the record for 40 LED's in a line.



Which is not to mention that it is tiny as a lightsource which doesn't match the measurement of the fixture listed. I'm still leaning towards a single LED...

Best,
Jason.

[fuso]

I'm still leaning towards a single LED...

I can only second that as I've done a test with two types of horizontal fins, dichroic and frosted glass. The first image
is a set-up with small spheres (no clones) using the same IES emitter assigned to each one of them. The second image
is using a long rectangular single-sided polygon which has an MXI based emitter assigned to it. The MXI is just b/w
where white dots represent roughly the size and spacing of the LED's. The only difference I can see is that it is brighter
and that the caustics are more defined. In terms of render time the IES version rendered a bit faster (against all odds)
You should look at the images in full resolution to pick up the detail and see the 'LED's'.




As for the dichroic glass, I have no idea why it renders so dark. It's ok using the daylight set-up but now being backlit it
won't brighten up even though it has reached SL 17. Maybe I have to let it cook for longer but in the render preview it
didn't look like it's gonna brighten up dramatically... So, back to square 1.

Regards,

Jost

[Half Life]

The IES version looks more realistic to my eyes -- I've noticed something about coatings (and I'm not positive about this), I think the ND of the coating overrides the ND of the glass in some way... I would be curious to try the coating with the same ND as the glass and see if that helped.

Also, I would move the light source a bit more back from the edge of the glass.

Best,
Jason.

[Half Life]

Yeah, I'm pretty sure (based on some tests that I did) that the coating overrides the Nd of the glass to certain extent... So for better results I'd say make sure the coating has the Nd of your Glass material.

Best,
Jason.

[jwiede]

Sorry, kind of distracted by the holiday. I definitely see what you mean about the IES domain itself, it doesn't look quite oblong enough. I'm still trying to wrap my head around how the different axis dimensions correlate to physical emitter (IES's aren't a normal 2D projection, distance along X or Y changes view angle), though, before I'm convinced it's wrong for a strip of them. I'm thinking that for the IES to look oblong at all indicates a major width-vs-height emitter surface difference in physical emitter dimensions. Will comment more after a bit more thought on it.

So do you agree that if it represents a strip, the emitter shape should be a cylinder or hemi+cylinder+hemi? Or should it still be a sphere?

(broken ND coating theory removed)