Dan Margulis Applied Color Theory - 5000K Lighting discussion

From: INTERNET:Kim_Oravecz@averydennison.com, INTERNET:Kim_Oravecz@averydennison.com
Date: Fri, Oct 22, 1999, 1:53 PM
Chris Murphy <lists@colorremedies.com> on 10/21/99 04:24:01 PM
Subject: Re: Question about RGB gamut and recommended reading?

Now for another exciting thread, I'll let someone else ask the list (and me) why fluorescent 5000K bulbs are not the best kind of lighting to usefor proofing and design work.

From: Richard Kenward, INTERNET:kenward@photoimaging.demon.co.uk
Date: Fri, Oct 22, 1999, 4:47 PM

May I just saw that when I am being very picky/uncertain, I will go to a north window and check there also. For some reason, I find that there is just something 'missing' from the fluorescent 5K bulbs:-)

From: Chris Murphy, INTERNET:lists@colorremedies.com
Date: Fri, Oct 22, 1999, 5:53 PM

It involves a concept called metamerism. Metamerism is when two samples (of color) appear to be the same color under one lighting condition, but not other lighting conditions. They are called metamers, or a metameric match. The reason why this happens is that the two samples of color are made up of different stuff (different pigments, dyes, whatever); so when these two different samples made of different stuff are placed under any given lighting condition; they absorb wavelengths of light in different ways. Sometimes the differences are minor so the color samples look the same. Sometimes the differences are large so they look different; and those differences change not only depending on the stuff the samples are made up of, but the lighting source too.

Some of you might be able to relate if you have had two pieces of clothing look the same under indoor incandescent lighting; but outdoors they looked really different. That's metamerism - two color samples look the same in one lighting condition, but not under other lighting conditions.

The high end proofs (expensive ones) usually have laminates made up of actual pigments used to make inks for printing presses. This means that while the laminate isn't a liquid ink, it has the same properties (spectral properties) as the inks used in a printing press. This means that these kinds of proofs, and press sheets are NOT metamers. No matter what lighting condition you view them under, they will both look the same (they very well may not look correct in terms of non-standard lighting; but BOTH will look the same).

However, with the increasing use of digital proofs that don't use actual press ink pigments, they aren't made of the same stuff. This means that an inkjet proof (for example) and a press sheet ARE metamers. This means that under non-standard lighting we can expect an inkjet print and press sheet to not look the same.

WHEW - lots of background there. So the real question is, what is standard lighting? Standard lighting in the United States (and increasingly so elsewhere) is D50, *NOT* 5000K. 5000K is just the

temperature where as the standard lighting also specifies the ILLUMINANT, not merely temperature.

What's the big deal? The spectral property of the light source depends on both the illuminant and the color temperature. There are bunches of illuminants, A for incandescent, B which is no longer used, C is tungsten simulating daylight, D is daylight (for which four temperatures exist), E for theoretical equal energy, and F for fluorescent. There are LOTS of F type illuminants, F2, F5, F6, F7, F8, F9, F11, F12 and others. Not all fluorescents are the same, as you've noticed when mixing and matching fluorescent tubes.

So what's the big deal with 5000K fluorescent not being good enough? It has a spectral power distribution that makes them far closer to the F2 or F7 illuminant than D50. Keep in mind D50 is the target, not F2 or F7.

Here's an analogy. Your kid comes home with a report card with an average grade for all classes. It says B+. Doesn't sound too bad right? Well if you were to look at each subject individually, you'd find he got an F in History and A's in everything else. You wouldn't be too happy with that F right?

Saying a bulb is 5000K is like an average grade. It doesn't tell you about specific areas it's good and bad at, that is, where the grade F is located.

There is only one bulb that I have come across so far that is the closest to simulating the D50 spectral power distribution (a fancy way of looking at ALL of the grades in all subjects for a light source). SoLux 4700K lamp. Don't be thrown by the 4700K rating. The important thing is the spectral power distribution, even if that sounds like geek speak.

www.soluxtli.com - it's a standard size MR-16 halogen bulb with a special filament and reflector. Naturally they've got patents on this thing. The price is right too. (I don't get free product, discount product or kick backs from this company; I just like the product and so far I haven't found a better product.)

So with this bulb you could turn an entire room or workstation area into a proofing booth, err proofing ROOM. Far better than fluorescent.

For those who are environmentally concerned, fluorescent bulbs (all of them) contain mercury. mercury is bad. Mercury is also the reason for the large spikes in the spectral power distribution. For photography, this typically causes photographic paper to have a pinkish or magenta cast to it; people look a bit sunburned. It's the lighting source, not your device, not your color management software, not your color correction.

Lighting is one of the biggest reasons for color problems. Now you know :O

From: Andrew Rodney, INTERNET:andrew@digitaldog.net
Date: Fri, Oct 22, 1999, 5:55 PM

Fluorescent bulbs can be rather "spiky" in their distribution over the spectral range. The SoLux bulbs are supposed to be quite good for light boxes. In fact, Joseph Holmes had a nice recipe for making a home made box using the Solux bulbs which he described (with part numbers and web pages) on the ColorSync list about 6 months ago. The title of the message was "Re: Tips for monitor calibration, SoLux" which is archived on the ColorSync list. I've been interested in trying out the recipe which doesn't cost that much ($200).

From: Chris Murphy, INTERNET:lists@colorremedies.com
Date: Fri, Oct 22, 1999, 6:46 PM

And you are not only correct about something missing, but they often have something in ADDITION (spiky emission at mercury emission lines usually) that makes them NOT D50.

Beware the north window! Over the course of the day, the color temperature of daylight varies from 4773K to something like 6540K.

Anyway, that can make a big difference. Also, some kinds of inkjets and solid ink printers use inks that are sensitive to UV. There is very very very little UV in D50; but A LOT of UV in sunlight. So beware :)

Long live SoLux! (Actually they are working on generation 2 which might include a 5000K bulb instead of 4700K which would make this bulb the ruler of the universe and would totally put all the fluorescent makers out of the 5000K bulb business. The 4700K bulb can do that now technically but people have this thing about numbers, they want 5000K even though that's the minor detail.)

From: Chris Murphy, INTERNET:lists@colorremedies.com
Date: Fri, Oct 22, 1999, 6:48 PM

I think they have a box using their bulbs that's premade also. Not sure what the price is. They aren't yet making a box for transparancies however which would be nice because emulsion has this tendency to absorb a lot of green wavelengths (which fluorescents really output strongly compared to SoLux or D50 for that matter), and reflect red wavelengths.

From: Chris Murphy, INTERNET:lists@colorremedies.com
Date: Sat, Oct 23, 1999, 12:54 AM

They all have a spiky power distribution which to me makes them less than ideal. They give off massive spikes at various wavelengths which inherently makes them non-D50 in their behavior.

If the goal is to standardize lighting on D50, then you need a lighting source that best matching that illuminant. Fluorescent doesn't (it can't really) have this ability.

> Try mixing 3250k with daylight for a pretty good method of "gut level" evaluation.

Daylight changes throughout the day. Not only that, but the window glass, trees outside, concrete on the ground, all of those affect behavior.

What is the delta E I'm talking about in terms of 5000K fluorescent vs. D50? Anywhere from 4 delta E's to more than 20 delta E's. 10 delta E's is unacceptable. The average person can see a difference of 5 delta E's; so for some even 5 delta E's is unacceptable.

I'm happy with the SoLux bulb, and it's far less expensive than other solutions I've come across, and with superior performance; and more practical from the standpoint of, I know it works and why would I want to borrow a spectroradiometer anyway; that would be expensive.

>Also, try/borrow a color temperature meter to really get a
>grasp of the problem. Remember, the human RGB neuron system has built in,
>on the fly, color correction.

It has chromatic adaptation but this applies primarily to whiteness. Fluorescent can alter the enter behavior of color, including gray balance and this is not something the human eye will adapt to.

A color temperature meter is insufficient. I can get an incandescent bulb with higher voltage (won't last more than an hour before it blows, but still) to make a color temperature meter to show 5000K. 5000K is NOT the standard. D50 is the standard. 5000K and D50 aren't the same thing.

A spectroradiometer is what you'd need to evaluate the suitability of a light source, and compare its spectral power distribution to that of D50. Spectral distribution is what's important. Color temperature is totally misleading.

From: Dan Margulis, INTERNET:76270.1033@compuserve.com
Date: Sun, Oct 24, 1999, 2:14 PM


Michael Stokes has been having problems posting to the list, probably due to a lack of tech support at his company. I'm therefore posting three messages he sent me.



> They all have a spiky power distribution which to me makes
> them less than
> ideal. They give off massive spikes at various wavelengths which
> inherently makes them non-D50 in their behavior.

Just for the curious, the spikes are due to the fact that fluorescent bulbs are based in mercury emissions energizing the phosphor coatings on the inside of the bulb which fluoresces at differing wavelengths. The light emanating from the bulb is a mixture of the pure mercury emission lines (extremely spiky) and the fluorescence from the phosphor coatings (fairly smooth). So fluorescent bulbs will always have some level of spikes due to their dependence upon mercury as a source.

Michael Stokes, Microsoft Corporation

If you,d like to participate with others in discussing Dan's book or related concepts, join the Applied Color Theory email discussion list!
Enter your email address below, then click the 'Join List' button:


Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.