red-green colored matricies colorblind friendly?

[jeffythedragonslayer]

There are some 16x16 mapper compatibility matrices on the wiki that seem to rely on color only to convey information. Example:

https://www.nesdev.org/wiki/Everdrive_N8_Pro

According to NIH, the most common type of colorblindness is red-green, of which there are 4 subtypes:

https://www.nei.nih.gov/learn-about-eye ... -blindness

Therefore I believe it would be an accessibility improvement to have color not be the only way to distinguish "yes" from "no" in these matrices. What does everyone think a good way to do that would be?

[calima]

Take a screenshot and convert it to grayscale. If the colors look different enough, r-g people will also see the difference.

Other common options are stippling, etc.

[lidnariq]

The standard advice for colors that can be perceived by all three types of colorblindness is to use orange vs cyan.

The problem with that is that people with normal color vision don't perceive those colors as having meaning, while red-vs-green is instilled culturally.

None-the-less, I've tweaked the shades of red and green to be slightly orange-er and slightly blue-er, which should still be perceived by trichromats as well as hopefully more nearly distinguishable by dichromats.

[jeffythedragonslayer]

Here is what the new colors look like converted with GIMP's Image->Mode->Grayscale:

Another thing we could do, that I think would be easily understood by both English and Japanese speakers (both languages associated with the NES/Famicom), is adding little "O" and "X" icons inside the boxes.

[Drag]

Some tips on UI:

If you put a color legend next to the table, you can clarify which color means which thing, so you don't have to rely on inference.

Next, prioritize different colors being different values (i.e., lighter or darker). Since this table is just two colors, as long as green is a lighter color and red is a darker color, everyone should be able to discern between the two. As a bonus, this also knocks out the edge case of someone wanting to print this out in grayscale, or someone using an e-ink display (grayscale again) to look at the wiki.

Finally, and most importantly, get feedback from an actual colorblind person, because they'll know better than us non-colorblind folk.


As a fun fact, in RGB, green is the lightest color, blue is the darkest color, and red is in the middle, and this is because of the way our eyes respond to different wavelengths.

[Bavi_H]

It looks like Chrome and Edge desktop browsers have color blindness simulation in their F12 Developer Tools:

1. Press F12 to open the Developer Tools.
2. In the ... menu, go to "More tools" and select "Rendering".
3. Scroll down to "Emulate vision deficiencies" and choose one of the options in the drop down box.

[Joe]

Take a screenshot and convert it to grayscale. If the colors look different enough, r-g people will also see the difference.

Different types of colorblindness have different impacts on perceived luminosity, so colors that appear significantly different when converted to grayscale may still be indistinguishable to a colorblind person. (And that assumes the grayscale mapping is based on perceived luminosity in the first place...)

It looks like Chrome and Edge desktop browsers have color blindness simulation in their F12 Developer Tools:

Firefox does too, there's a "simulate" selector in the "accessibility" tab.

[jeffythedragonslayer]

I put a legend in there for now, because it was easier than putting an image in every cell. I kind of want to wait for more feedback from a colorblind person before copying the legend to the other matrices though.

[Drag]

Different types of colorblindness have different impacts on perceived luminosity, so colors that appear significantly different when converted to grayscale may still be indistinguishable to a colorblind person. (And that assumes the grayscale mapping is based on perceived luminosity in the first place...)

That makes sense, the four types of cells in the human eye have their own response curves. The rod cells are sensitive to green-blue (but produce no color perception, just lightness), so if you're missing the M-cones, the rods could pick up the slack and you'd at least still see a similar brightness, but if you're missing the L-cones, then nothing in your eye is sensitive to the red region, so those wavelengths would appear much darker (as in, closer to black), not just duller (as in, desaturated), and same with S-cones and the blue-violet region.

I never thought about this, but if true, then a whole bunch of simulations and colorwheels are just flat-out wrong, so I'm wondering why this isn't more common knowledge? And yeah, that would imply that the conversion to grayscale would be different for each kind of color blindness.

[calima]

It's the rarity of those deficiencies, I believe. Red-green is the most common colorblindness, the other types much more rare. Diminishing returns, what is good enough, etc.

[krzysiobal]

What about blind people? How a text-to-speech tool will read that table?

[Drag]

In that case, it'd be helpful to get input from someone who needs to use a screen reader and does NES development, if they could provide some comments on what can be done to improve huge matrix-style tables like this.

...and also what it's like to work on retro game console homebrews with such a visual impairment, because that sounds pretty interesting.

[Pokun]

It would be interesting to know if there are a lot of blind people playing video games at all.

[tepples]

Audio games: where every playthrough is a blind playthrough.

See "The 7 best audio games" by Milagros Costabel

[Pokun]

That's pretty cool and good to know in case you ever loose your sight.
I suppose any text-only game and games with minimal graphics like the original Wizardy could work well as an audio game.