RE: Please vote on proposed WCAG 2.1 SC for Color Contrast of User Interface Components

Responses below…

Steve


> I do not subscribe to the logic that text is always much harder, and using that to oppose the SC is just as futile since there is no research to back it up either.

There isn’t any research the other way either, that I’ve seen.
[Steve] Yes, my point was just that arguments for or against need to be based on either common knowledge or logic, or research.  Saying text is harder doesn’t pass the former.

NB: I’m not opposing it, I’m trying to deal with comments from Gregg and others on the Graphics Contrast SC. Currently I don’t have a repost.
[Steve] I understand – I did not mean you personally.  Sorry if my email came off like that.


>I can say interpreting complicated plots with several line and/or symbol types or busy diagrams is certainly not much easier than reading text.

I’m sure that has extra cognitive load, but I’m less sure it has the same level of acuity needed?  They are generally thicker and better spaced out than text (even on maps and circuit diagrams).
[Steve] I’m just going to have to disagree.  I strain my eyes all the time viewing things like this, and a map with tons of lines and labels is actually a perfect example.

Perhaps not all the time, but it has be balanced against the need for multiple colours, as using 4.5:1 with multiple colours is far harder than with text on a background.
[Steve] I would not be opposed to a drop to 3:1, but why is it so hard?  I think the comments about not being able to use many colors misinterprets the actual criterion.  On a bar chart, for example, a bar may need to be distinguished from 3 adjacent colors (2 bars on either side and the background).  A thin border with good contrast to the background solves the problem and then any color for the bars passes.  Where the SC is currently deficient is when objects do not have a solid color and when there are many interactions between objects.  For those, we need to figure out a way to say that the dominant colors need to contrast (e.g. worry about lines contrasting with the background and not when they cross each other).


> graphics often contain a lot of text anyway (e.g. scales, labels, legends, etc.).  These aren’t covered by WCAG 2.0 because the definition for “images of text” excludes images with other significant visual content.

Interesting, a diagram with text in it would avoid 1.4.5 images of text, but a diagram with separate images for each label would fail under that SC.

Maybe we should extend graphics contrast to images of text? E.g.
“The visual presentation of graphical objects and images of text that are essential for understanding…”
[Steve] I think in this case the text has to fit the graphical object definition.  We can’t say images of text because that would circulate back to the glossary which excludes what we want.


> consider color blindness.  For these users, it’s not about following anything; it’s just about seeing it.  If I make a very simple icon using two conflicting hues with the same luminance, what accessible path do they take to see it?

Luminance is the basis for the contast check, so if the whole icon is a ‘graphical object’, then the outer part of that graphic are what is required to differentiate it from the surrounding (both colours against the background, but not between each other).

If you need to distinguish the colours within the icon to understand it, then each is a graphical object and it should distinguish against each other and the background (assuming they are both adjacent to the background).
[Steve] Yes, this is the case I’m referring to.  My point was just that color blind users benefit from contract criteria, so the opposing argument about reading text falls short there.

Cheers,

-Alastair