How accurate are EOS's filter colour presets?

Amy Worrall

Member
I'm playing with ETC Nomad, and took a look at the filter library. Some of the colours seem rather odd to me.

For example, L061 "cool mist", which I know as a very subtle cool wash. When patched to a generic LED PAR, EOS seems to suggest it as R100, G94, B68, which to me seems far too warm for L061.

Likewise L117, "steel blue". I know this one as a fairly light cool wash with a touch of green in it. EOS gives it R82, G100, B77. I'd have thought there would be far less red in it than that.

I know that the values are never going to be perfect without calibrating to a specific light, but these seem way off. Can anyone reassure me that EOS knows what it is talking about?
 
The color presets vary wildly between fixtures. Generic personalities are especially erratic. There are reasons for it, are you used to seeing the L061 in a S4, a PAR64 or a 6x9? It's going to look different in those three so to which are we matching? As for the RGB values for L117, while blue it is a very light, unsaturated blue. If white is RGB all at 100% then this might be close. Best way to find out is to hook up a light. Remember that we are talking additive color mixing to get to a value we know from subtractive. That being said the presets for the ETC fixtures are usually pretty close and other manufactures are in the ballpark normally. I find that I use the presets as starting points. In some cases they work for the job, in others I pull out a S4 and the gel I am trying to match, point that and the LED I am working with at the cyc and create my own pallet.
 
ETC does the the best it can with the information available but sometimes gets it wrong. The Carillon fixture library does not have any calibration data for generic fixtures so the results are going to be dubious.
 
Then you have to consider your monitor. It all gets dumbed down to video RGB and compared to the lighting in your room. Color matching gets complex in the best situations!
 
It's not just the on-monitor representation though, I was genuinely surprised that a cool colour like L061 would have more red than blue in it.

As soon as I'm next home (in a week or two… lots of travelling at the moment!) I'll throw up a fresnel and an LED PAR and compare.
 
Another point I haven't quite seen anyone make yet:

R100 G100 B100 isn't white on any RGB PAR I've seen yet. You generally have to back 2 of the colors off to get to white-ish; which two probably depends...
 
Different emitters vary in output.

White light is not necessarily equal amounts of all colors.

Eyes vary across the spectrum as well.

Color perception depends on the visual environment.
 
While I understand the motivation, I personally curse the entire gel picker section of the console. It leads to too many confusions like this one. Whenever I do board training, I try to steer programmers away from there, and towards straight mixing. Even there, two fixtures won't match, but I've found people are more relaxed about them not matching when they've mixed. Designers have a very firm look in their mind when they ask for R26, but not as much when they ask for R100, G020, B050.

The lime enabled fixtures generally have a pretty good base white with everything at full. The colorsources are tuned towards a warm white. Some others I've seen have been biased towards a cool white.
 
We just got this feature on Avo Titan software and it's pretty hit and miss. Not a big deal for me. It's nice if I can just hit a color and get close then tweak it manually from there. And then there are the stupid ones like if I pick R27 then it just brings up the red to 37%, which is just plain stupid IMO. So I'm sure there are a lot of other colors where you could end up with a lot more output if you just manually mixed the color you wanted.
 
I would like to say, the console manufacturers have put a stupendous amount of work into making this feature remotely useable. The problems come from trying to treat modern lights like conventionals, and the language starts to breakdown. Thanks Rob for that article. It has been a few years since I read that one, it is a good read.
 
Apple solved this problem for computer hardware with ColorSync. I wonder if there would be takeup of a hardware system for calibrating lights (a calibrated digital camera plus a white reflective board, plus clever software that measures the light’s response to different RGB values)?

We’d need a console manufacturer on board, but it wouldn’t need the fixture manufacturers to do anything if the calibration routine was doable by third parties.
 
Apple solved this problem for computer hardware with ColorSync. I wonder if there would be takeup of a hardware system for calibrating lights (a calibrated digital camera plus a white reflective board, plus clever software that measures the light’s response to different RGB values)?

We’d need a console manufacturer on board, but it wouldn’t need the fixture manufacturers to do anything if the calibration routine was doable by third parties.

That gets you part way there, and probably closer to how it is now. But the problem is that different fixtures have different LEDs in them. They don't always match even in the same model of fixture, let alone in different brands. ETC and a couple others that do a really good job at binning the LEDs and keeping them all really close could tune each model to closely match a gel color. Then that would have to be built into the fixture profile, and the software of each brand of console would have to change their software to deal with that. But once you get into lower end "professional" fixtures, DJ fixtures, and especially generic Chinese fixtures then all bets are off. The binning isn't nearly as good, and the LEDs colors will shift over the time of the production. I could maybe see ETC doing something for their consoles and their lights, but beyond that I doubt anyone would bother making that much effort. And as time goes on more and more people will only be working with LED and arc source fixtures and not really know gel colors like those of us who have been doing it for a while. For those of us who are familiar with gel colors it's easy for us to think I want R21 or L201 or whatever and hit that color button. But for someone who doesn't really know their swatch books it's going to be much faster to just use a color picker or manually mix the color they want. So while I would like to see it happen, I kind out doubt it will. It would make life a lot easier for tours using house plots if they had a way to quickly and easily get the exact color they wanted out of a road houses fixtures.
 
For every complex problem there is an answer that is clear, simple, and wrong. -- H. L. Mencken

This is a a very complex problem. There are physical limits to using a random collection of narrow-band emitters to produce exactly the same colour as a broad spectrum source. Colour balancing a computer monitor versus a sheet of paper is a significantly more controlled experiment than colour balancing a random assortment of set pieces, paints, fabrics, skin tones and other surfaces on a stage.

Imagine playing a song on a piano with some of the keys missing. It might be possible to approximate the composer's intentions but an exact match would be exceedingly difficult.
 
you'd be lucky to get an rgb fitting to give you a CRI of 80 which means 20% of the colours are distorted or missing so no colour picker is going to be more than a rough approximation.
 

Users who are viewing this thread

Back