[mpetroff]

The simulation technique [1] currently used by the Firefox / Chrome dev tools is reasonable, so it's definitely much better than not checking for color vision deficiency accessibility at all. Ideally, color should only be used for progressive enhancement; if your UI can be used in grayscale, it's probably fine.

However, the method originally used by both browsers, which is also used by some other tools, had no scientific provenance and produced clearly incorrect results, so one needs to be careful about this sort of thing in general.

[1] https://doi.org/10.1109/TVCG.2009.113

[convolvatron]

colorblind person. making sure things look ok in greyscale is excellent advice. for me perceptual difference in intensity are most of what my visual system pays attention to...color only if I try.

most of the the colored interface are hugely garish, and the massive intensity differences actually make it really hard to read.

and conversely if two indicator colors are the same brightness and you need me to distinguish between them, we're going to struggle.

[danShumway]

One thing I worry about here that I should probably do more research on: is colorblindness like turning off or lessening a color channel? Because I know for some graphical work I've done, if I do a pure grayscale by just turning down the saturation or something, the contrast might be good, but if I turn off one color channel, the contrast/intensity differences sometimes seem to vanish.

I do grayscale testing, but I worry that when I do grayscale tests for stuff like game graphics where there aren't clear contour lines, that I'm getting a false sense of confidence -- I worry that the test might not be an accurate representation of the level of contrast a colorblind person will actually see if they sit down in front of the non-grayscale version. Is that something I should be worried about, or am I just misunderstanding how colorblindness works?

If I do a grayscale test and the interface still seems clear/usable, how confident should I actually be that you'll be able to use the non-grayscale interface?

[convolvatron]

part of the issue here is while the physical mechanism of colorblindness is quite clear, the perceptual implications are inherently subjective.

so I don't know how much my advice is relevant to colorblind people in general.

but greyscale is very good start. after that - if you want to show contrast at least for me there are whole areas of the palette that you don't want to take more than one color from. for example the whole red/green/brown/tan/orange space for me is a giant wash. purple looks nice, but don't set it against blue unless its substantially different in saturation.

a secondary issue is that some color combinations I find confusing/distracting. that is my brain can't really settle on the color(s). so for example casual games sometimes just don't work, because its going to take me a second or two to even decide if I can combine these two dots...and I may not even be able to focus on anything else because there is the loud flashing thing on the screen (flashing because I'm 'trying on' colors to see if they fit)

if there is too much color going on I'll probably just reject the whole activity if I can.

sadly I've seen games with a 'colorblind mode' which shows that someone really cared - but it was no more usable than the default more or even worse.

but low saturation UIs with good color choices do look better than mono. I have some form of blue/yellow, but certain choices of blue and yellow are really quite readable and even look pleasant (shrug).

but I wouldn't flail too much. as other people have mentioned its not really that life-changing. if I have to use your garish interface to book an appointment it doesn't really matter...but if you're going to go in and colorize my emacs with nice round hex values I wish you would have the courage to post your address so I can come visit you and have a talk.

that was all over the place. hope it helped.

[danShumway]

That is super helpful. I've done a little bit of research on this, but apparently not nearly enough because I haven't really thought that much before about the distraction angle at all; I assumed if I got the contrast right for everyone, that there just wouldn't be any other problems with color.

I guess this is a good argument for not just testing in grayscale, but at least thinking about actually shipping the grayscale mode or individual color channel adjusters. Because I know in a worst case scenario that a grayscale test will be accurate to what people see if I provide a toggle that turns it on in the final product.

Just... yeah, to your point, need to make sure that whatever colorblind mode is shipped is actually an improvement :)

[arthur2e5]

What was the original method may I ask? Some… basic RGB mixing?

[mpetroff]

It applied a linear transform to RGB values, like the current method, but the original matrices were the "colorjack.com matrices" [1] that have been floating around the internet for a while. They weren't derived scientifically and are quite inaccurate, at least for protanopia.

[1] https://github.com/MaPePeR/jsColorblindSimulator#the-colorma...