Wednesday, December 12, 2012

Follow up on colorblindness testing

I posted a blog on colorblindness testing (and apps for tablets, and gamuts of displays, and my dog Truffle) that has led to some interesting discussions in various LinkedIn groups. The comments are great, but unfortunately, they wind up in a variety of different groups. People from one group don't necessarily see what people in another group have had to say.

In this blog post, I am coalescing the various discussions.

How can you test for colorblindness?

The one true standard test for colorblindness is the Ishahara test. The picture below shows slide 4 from this test. People with normal color vision will be able to clearly see a picture of Eddie Albert [1] in this picture. This was developed back in 1917 by an ophthalmologist whose name was (oddly enough) Dr. Ishahara. The test is quick and easy to perform, and you can find 867,324 versions of this test online, although I did notice that there are a few that are not quite as good as the other 867,319.

A picture of my Academy Award winning spaghetti

The other one true standard for colorblindness is the Farnsworth Munsell 100 hue test, which, according to the XRite website is "one of the most widely used tests in industries where color decisions are critical." You can by the boxed set from various places including "the Munsell store". I am not sure how they came up with that name. Just another fluke, I guess.

The Farnsworth Munsell 100 Hue test with 84 hues

In the test, the contestant is given one rack of tiles at a time, and is asked to put them in order according to hue. This is definitely a tough test - the difference in hue between adjacent tiles is very subtle. For that reason, there is some substantiation for the claim of the Munsell Store that this is an important test for people who are making judgments on color matches where the color is critical.

David A. (one smart guy I know) pointed out the difference between the two true standards for colorblindness: "As the Ishahara test has fewer testing points than the FM test, the FM test is also a great test to test for subtle discrimination..." David is making a distinction here between colorblindness and something more like "color acuity". I have seen myself that there are (loosely speaking) three types of people when it comes to the FM test: 1. those who fail miserably and are truly colorblind, 2. those who do very well on the test, putting almost all of the colors in the proper order, and 3. those who get a bunch wrong. I honestly don't know what distinguishes the second and third categories.

Paul raised this point in a way that sounds like he is pretty smart, too: "to find out who in the staff that has perfect colour discrimination capacity (roughly only 10% of the male and female part of the staff), you need to use the FM100 test (or similar)." [2]

Two other people (separate LinkedIn discussions) Shoshana and Michio provided links to the online version of the Farnsworth Munsell test. This is certainly cheaper than the physical version, and it is part of the webiste from XRite, so you know it's not from some fly-by-night guy who claims to be a color scientist and maybe has a silly blog where he pretends to know what he is talking about.

Screenshot for the XRite FM test, which has nothing to do with radio

But this might not be a good test. Here is a comment from that European (and therefor presumably smart) guy, Paul: "One of the teachers at Malmoe University, Graphic Arts Department, tested the online version of the X-Rite (simplified) FM-100 test, and failed miserably. I pointed out to him that he probably needed a better quality monitor, and a calibrated one. I suggested he try my monitor, a LaCie at the time, which was fairly recently calibrated. Now he suddenly managed very well in the same test."

I tried out the online test myself. My score was not all that spectacular. I appreciate the fact that Paul has offered me a way to save face. My monitor is not calibrated.

This brings up the next discussion topic...

Are the tablets up to the task of accurate color testing?

Here is the seed that got me started on this blog post. Bob C. apprised the group "ISO TC 130, WG 13, Task force Harmonizing" of the EnChroma app. Bob was wondering the basic question as well: "I'm also interested to know if EnChroma test correlates with the Ishihara's test for color blindness."

Here is my opinion, for what it's worth. I think the EnChroma app is likely to work well to screen for colorblindness.

David A. asked a similar question in a different thread: "I wonder if the three tablets tested would show any significant differences with an FM type of test."

It would appear that we have two pieces of anecdotal evidence that suggest that color management is important, at least for the FM test on the display of a "real" computer. This makes sense. Colors change when I move a window from my laptop to the monitor that it drives. It is not that far of a stretch to say that the difference between two colors can change as well. The display on my laptop compresses some areas of color space. If the compression happens in areas that the FM test is looking for, then the test will just get harder.

This in turn, brings out the next question...

What about color management on tablets?

Reem pointed out that the Datacolor SpyderGALLERY is an app for iPad and iPhone that provide some color management solutions for photographers. This gadget will measure the colors on your iPad and make adjustments so that it will provide accurate color rendition of images.
The Sypder4 monitor calibration device

She asked if there were any similar apps available for Android devices. I am not aware of any. Sounds like an opportunity for some company... Maybe Datacolor or XRite have some interest in this?  Or maybe the Android operating system does not facilitate calibration?

Does anyone have an answer to this?

One last thing

For those unfortunate readers who are not from the Midwest, I may need to clarify a subtle joke in the previous blog. I called the folks who took the test "Ollie", and "Lena". I should have called the fellow "Ole". Steve T corrected my spelling on this.

The names came from one of the favorite pastimes of Midwesterners: telling Ole and Lena jokes. Ole and Lena were an old Norwegian couple, and they were always getting a bit confused. I wrote these two joke sespecially for this occasion.

Ole: "De doc says dat I am color blind to vun color."
Lena: "Vut color is dat?"
Ole: "I don't know. I couldn't see it."

Sven noticed his old friend, Ole eating a bag of rabbit turds.
"Vat are you eating der, my friend?"
"Lena gave me a bag of M&Ms for my lunch. Wasn't dat sveet of her?"
"Ole, didn't you notice anything funny about de color?"
"Ach, yes! Lena always takes out all de red ones cuz knows I'm color blind. Such a dear voman she iss"

[1] Eddie Albert played the itinerant peddler Ali Hakim in the movie version of Oklahoma, and was nominated for Academy Awards in two other films. After sunc accomplishments, how would you like to be remembered for Green Acres? Life is not supposed to be fair.

[2] Another clue that Paul is smart is that he puts the "u" in "colour". That means that he is either European or pretentious. Since I know that he is European, and all Europeans are smart, then... I don't think that David A. is from Europe, although I would guess that he has been there a few times.


  1. Wow, that was a surprise! I took the online FM color test and got a perfect score (I saved screen shots as proof). I've never thought my color perception was very good. I'm terrible when it comes to matching colors from memory. By the way, thanks for acknowledging the whole "Ollie/Ole" fiasco. I'm sensitive to it because my maternal, Norwegian-born grandfather's name was Ole.

  2. So I go online to the X-Rite site, and search on online color challenge, no hits. Have they removed it to sell more FM Kits?

  3. Congrats on the perfect score, Steve! (mutter, mutter, mutter under my breath... he must have a much better display than I have... I am sure he used a color picker tool to yank RGB values out of the image... As for the screen shots, I'm sure he is capable of downloading the HTML and putting in any score he wants...)

    You should be proud to have done so much better than I did!!!

  4. Hutch,

    Sorry, the link to XRite's online test was buried as a link:

    It does seem odd that they are "giving the test away" online, when they make a bit of change selling the FM kits. (I am sure the kits are expensive to make, so I don't think the price is ridiculous.)

  5. John,
    I have been meaning to jump in this for some time but have been pressed for time lately. The Ishihara plates are not the definitive test for colr vision deficiency - the Nagel Anomoloscope is. The Ishihara test is an example of a pseuoisochromatic plates. There are or at least were some competitive plates. The oldest that i have seen were sold by American Optical (same folks who make those really cool instruments in your optometrist's office. Those plates were designed by Hardy, Rand and Rittler so were known as the AOHRR plates (test). The were different in that they used geometric shapes ( circle, square, triangle) instead of numbers. They also had some supplementary plates with wiggley lines so the subject could simply trace the "spaghetti" across the plate. By the way, Hardy mentioned here was a professor at MIT who designed and built the first color measuring spectrophotometer ad wrote the first textbook on colorimetry, describing the newly published method of colorimetry recommended by the CIE.

  6. Thank you, Unknown, for the information. You know that I just put a Nagel Anomoloscope on my Christmas list!!

  7. I liked the Norwegian jokes (we Swedes love jokes on especially Danes and Norwegians).

    An example: A Norwegian did his preflight check on the car before starting off, like every responsible driver does. But he got stuck when he came to the first indicator lamp: Works, doesn't work. Works, doesn't work. Works, doesn't work. Works, doesn't work. Works, doesn't work. Works, doesn't work. . .

  8. As a half Norwegian, I can attest that my indicator light does the same thing!

    Thanks for the good joke.

  9. Let me echo the earlier comment: the "one true colourblindness test" (yes I'm European) is the anomaloscope: mix red and green lights to match a yellow stimulus.
    I have been teaching color science for decades, and my view of the FM 100 test is coloured (!) by the deuteranope who scored perfectly on that test: it is a far better acuity test than it is a CVD (colour vision deficiency) test.

  10. Finally got around to posting my response to your response to my post...
    Thought it might be interesting to really check the accuracy of some of these devices so I plotted Gretag Macbeth color checker colors as reproduced by an iPad mini and Nexus 7. The two devices reproduce this set of colors quite differently. Its interesting that this did not seem to affect your initial test results:

  11. Martin, thanks for the further information. That's strange that someone who has colorblindness could score perfect, despite the fact that most people with normal color vision do not. It seems as though CVD is "in the eye of the beholder", while color acuity is in the brain.

    Jeff, excellent post, thanks for the followup. My explanation for the test working on a device with limited gamut is that, even though colors are distorted compared to what they are "supposed to be", the relative placement of colors is not changed much.

  12. I've taken the x-rite test a few times and generally find it pretty easy to score perfectly. But then I color correct images almost every day. The monitor I use has settings for emulating various color blindness types and Martin's comment made me wonder how the test would play with the monitor set on deuteranopia emulation.

    The answer is: it's hard. I can image that a one or two of the lines might be doable if you were used to seeing the world this way. But the chips in the third line are all very similar in color making it, for me, impossible to even begin.

  13. A very useful blog. I am very much benefited after reading this post. Keep sharing.

  14. A couple of random comments:

    1) It seems a bit suspicious to compare an on-line hue discrimination test based on mixtures of RGB primaries to one based on pigments that might actually have a spectral peak in, say, the yellow or cyan region of the spectrum. This bears some relation to the idea of using more than three primaries for color reproduction, so that people with variations in color vision see something closer to what they see with the original scene spectrum.

    Has anyone thought of devising a three primary, two match point (yellow and cyan) anomaloscope, to test the discrimination of women who have four cone cell variants? This could settle the question of whether their color space is truly four-dimensional.