Good luck documenting the diferences between the different ways to convert an image to grayscale:

https://graphicdesign.stackexchange.com/questions/69308/how-to-convert-color-images-to-black-white-in-gimp/69372#69372

On 6 June 2017 at 06:02, Lancer wrote:

I am a school teacher. One of the checks I ask students to do in order to test the contrast of their graphics work, is to convert the images to grayscale and see whether images are still clear.

There are two methods students are using to convert their images to grayscale for this test...

Method 1: flatten image, then Colors > Hue-Saturation => slide the saturation slider down to zero.
Method 2: image => mode => grayscale

Either of these methods results in a grayscale image, but the grays are not exactly the same.

For example, if I have absolute red (#FF0000) next to blue, the grayscaled-blue may match the grayscaled-red depending on the tone *and* the method used. Method 1: Absolute red (#FF0000) will grayscale-match absolute blue (#0000FF) Method 2: Absolute red (#FF0000) will grayscale-match a slightly lighter shade of blue (#2626FF)

Why are the two methods of grayscale having a different result? I would have thought that conversion to grayscale would be the same process as dragging down the saturation of an image.

...and given that they are different, which is the better method to use in terms of testing for contrast in media assignments?

-- Lancer (via www.gimpusers.com/forums) _______________________________________________ gimp-user-list mailing list
List address: gimp-user-list@gnome.org List membership: https://mail.gnome.org/mailman/listinfo/gimp-user-list List archives: https://mail.gnome.org/archives/gimp-user-list

I'll take a stab at this, but other more knowledgeable people may weigh in: as I understand color theory, the notion that there is a canonical equivalent grayscale value for every color is somewhat of a fallacy; any such translation relies on pyscho-perceptual assumptions about how we perceive different colors. As a result, translating from a color image to a B&W image is somewhat of an art (heavily informed by lots of research).

I'm not intimately familiar with the various algorithmic options in GIMP to do this conversion, but different options use different methods, as you seem to have found. I'm guessing, but: sliding the saturation to zero likely applies a standard transformation to go from RGB to HSV, then sets "V" to zero, and then translates back to RGB. This may involve different assumptions and techniques than "mode -> grayscale" which might do a more (or less) subtle transformation.

Here's another detailed link on the subject: https://patdavid.net/2012/11/getting-around-in-gimp-black-and-white.html

I'm fond of the "C2G" GEGL filter (which in the latest GIMP betas is under Colors -> Desaturate -> Color to Gray, but in older versions is apparently under Tools -> GEGL Operation (see the tutorial)), but it does a bit more than just convert the colors.

Also, a minor pedantic note: Colors -> Hue-Saturation -> slide-saturation-to-zero doesn't technically change the image to a grayscale image, as it's still an RGB image internally. And other exotic methods of making an image "black and white" might actually preserve some imperceptible amount of color information in the pixels to achieve the effect. (I think I've read about that, but can't cite anything off hand.)

-c

On 06/05/2017 09:02 PM, Lancer wrote:

I am a school teacher. One of the checks I ask students to do in order to test the contrast of their graphics work, is to convert the images to grayscale and see whether images are still clear.

There are two methods students are using to convert their images to grayscale for this test...

Method 1: flatten image, then Colors > Hue-Saturation => slide the saturation slider down to zero.
Method 2: image => mode => grayscale

Either of these methods results in a grayscale image, but the grays are not exactly the same.

For example, if I have absolute red (#FF0000) next to blue, the grayscaled-blue may match the grayscaled-red depending on the tone *and* the method used. Method 1: Absolute red (#FF0000) will grayscale-match absolute blue (#0000FF) Method 2: Absolute red (#FF0000) will grayscale-match a slightly lighter shade of blue (#2626FF)

Why are the two methods of grayscale having a different result? I would have thought that conversion to grayscale would be the same process as dragging down the saturation of an image.

...and given that they are different, which is the better method to use in terms of testing for contrast in media assignments?

For a quick & dirty contrast test I would get them in the habit of converting to greyscale. This converts the colour space from RGB (or CMYK) to Greyscale...but it may not give them the best perceptual rendering of a colour scene in B&W (greyscale) if what they **want** to end up with is a greyscale "B&W" image.

Rick S.

On Tue, 2017-06-06 at 06:02 +0200, Lancer wrote:

I am a school teacher. One of the checks I ask students to do in order to test the contrast of their graphics work, is to convert the images to grayscale and see whether images are still clear.

As you discovered, there are different ways to convert to greys. However, people's colour perception varies, with more than one in 10 having some form of "colour blindness" (depending on how you measure) or eye difficulty - and more than that percentage unable to read small text, of course.

I don't know of any accessibility checkers for GIMP; there are PhotoShop plugins. It'd be a good Google Summer of Code project I suppose, if that's still going. I might even be able to drum up some funding for work in the area, and/or technical resources.

Because of the differences in people's vision, I don't think it matters which method is used to convert. The people who have poor colour vision will be exactly the ones who see the brightnesses differently, e.g. if their eye doesn't respond well to reds (the most common problem with human males) then reds will likely appear darker to them. My father couldn't tell the difference between a traffic light that was all dark and one with just red showing. So all the methods will be "wrong".

If the goal is just to make sure the image reproduces OK on a black- and-white printer, have them send the RGB image to the laser printer, then convert to greyscale in several different ways and compare, and they can learn a lot, it's a good exercise. The default dot screen on PostScript printers is actually fairly mediocre and most professional graphic design software replaces it, or used to.

Numerically, it's about colour spaces and gamma and precision and the purpose ("intent") of the conversion. The gegl c-to-g filter sometimes gets much better results than either the mono mixer or desaturating.

Liam

Liam R E Quin 
ankh on irc
Web slave for fromoldbooks.org

You may try Colors -> Component -> Decompose, then select LAB space, BW image is the L, thats a lot better than anything else I have tried

Regards

-----Original Message----- From: Lancer
To: gimp-user-list@gnome.org
Cc: notifications@gimpusers.com
Date: Tue, 06 Jun 2017 06:02:29 +0200 Subject: [Gimp-user] When black and white is not black and white

I am a school teacher. One of the checks I ask students to do in order to test
the contrast of their graphics work, is to convert the images to grayscale and
see whether images are still clear.

There are two methods students are using to convert their images to grayscale
for this test...

Method 1: flatten image, then Colors > Hue-Saturation => slide the saturation
slider down to zero.
Method 2: image => mode => grayscale

Either of these methods results in a grayscale image, but the grays are not exactly the same.

For example, if I have absolute red (#FF0000) next to blue, the grayscaled-blue
may match the grayscaled-red depending on the tone *and* the method used. Method 1: Absolute red (#FF0000) will grayscale-match absolute blue (#0000FF)
Method 2: Absolute red (#FF0000) will grayscale-match a slightly lighter shade
of blue (#2626FF)

Why are the two methods of grayscale having a different result? I would have thought that conversion to grayscale would be the same process as dragging down
the saturation of an image.

...and given that they are different, which is the better method to use in terms
of testing for contrast in media assignments?

-- Lancer (via www.gimpusers.com/forums)

An interesting test. I believe your test image has to be flattened to work?

Cheers, Rick S.

-----Original Message----- From: Lancer
Sent: Wednesday, June 07, 2017 8:13 PM To: gimp-user-list@gnome.org
Cc: notifications@gimpusers.com
Subject: [Gimp-user] When black and white is not black and white

Some very interesting responses here, thank you :-)

This would be interesting material for students wanting an extra study at a more
advanced level beyond level 1 NCEA; simply creating a poster or brochure with
"good" design principles (contrast, alignment, repetition, proximity etc).

It would make a good topic for Level 3, having a student analyse the different
hex levels of the grayscale conversion methods and to try and reverse engineer
the algorithms which may have been used.

I'm attaching a gimp image in color which "detects" which method is used when
turning it to grayscale. (Could be used to demonstrate to students that grayscale conversion is not always the same). I was entertaining spending more
time with the image, perhaps making a version where there is blue snow that make
the writing invisible until grayscale is applied. (Just a bit busy with marking
and lesson planning at the moment)

But thank you for your replies - great information.

Attachments: *
http://www.gimpusers.com/system/attachments/603/original/TEST_you_have_used_01.xcf

Lancer (via www.gimpusers.com/forums)

Liam R E Quin writes:

I don't know of any accessibility checkers for GIMP; there are PhotoShop plugins. It'd be a good Google Summer of Code project I suppose, if that's still going. I might even be able to drum up some funding for work in the area, and/or technical resources.

There's View->Display Filters, "Color Deficient Vision".

That's just one type, and there are lots of different variants of color vision. I think I've seen other GIMP color filters but don't have a specific reference, but a web search of GIMP color-blind
gets some hits that might lead to more filters.

Because of the differences in people's vision, I don't think it matters which method is used to convert.

+1. It's amazing how much color vision varies among people. We think "green" is an obvious concept that means the same thing to everyone, but even among people with "normal" vision, color perception varies tremendously.

And think of Ansel Adams and all his work in the darkroom. There is no one "correct" black and white version of a scene; the art is in creating the one that shows what you want to show.

...Akkana