Re: contrast for images of text

If I understand you correctly, the programmatic determination (role, alt,
etc.) would satisfy 1.4.1. I am not sure of that. Since 1.4.1 is intended
among others for visually impaired users, then appropriate markup might not
be sufficient. Perhaps a visual cue is needed to augment the use of color,
as implied by the techniques for 1.4.1



On Wed, Apr 30, 2014 at 12:01 PM, Ramón Corominas
<rcorominas@technosite.es>wrote:

> Adam wrote:
>
>  So then images of text would fail in A which I believe
>>
> > to be reasonable.
>
> I assume that you refer to "images of text with poor contrast". I agree
> that they would fail 1.4.1 since, for low vision users, the embedded text
> is conveying the information relying on color alone. Even if the text is
> provided in an alt text, for example, many low vision users don't rely on
> screen readers and therefore might miss the information completely.
>
> Indeed, I see the technique regarding the "distinguisable links" as a
> logical interpretation of SC 1.4.1 that implicitly recognises the
> importance of contrast. Being a "sufficient technique" for SC 1.4.1, it
> seems to suggest that the contrary would be a failure, although no failure
> exists to indicate that. Moreover, the technique does not really refer to
> color, but contrast ratio between links and the surrounding text, which
> makes me think that it should apply to 1.4.3 and not 1.4.1. In fact, the
> role information is there, so color is obviously not the only way to convey
> the presence of the link. That said, I think that the technique is properly
> attached to 1.4.1, because it correctly assumes that not providing contrast
> is equivalent to relying on colour.
>
> I can even imagine a weird situation where the level A is theoretically
> met, although in practice there is a severe barrier for many users.
>
> Imagine the following image of text that acts as a heading for a link
> "more info":
>
> - Text: "Special offer! WiFi router for only $5.99"
>
> - Color: light gray (#CCCCCC) over white
>
> - Fails 1.4.3 AA (contrast ratio = 1.6:1) and 1.4.5 AA (image of text),
> maybe fails 1.4.4 also, although it does not refer to "images of text". In
> any case, it is also AA.
>
> - "Pass" 1.3.1, 1.4.1 and 4.1.2 due to the programmatically determinable
> text included in the @alt, apparently no more "A" criteria involved)
>
> Below this image of text we have a link "more info" in normal text, with a
> gray Color (#717171) over the overall white background.
>
> - Pass 1.4.1 (the information of the link itself "more info" does not rely
> on color, but anyway the contrast ratio with the "surrounding" text is
> 3.0:1)
>
> - Pass 2.4.4 (level A), since the "context" is provided with the image of
> text (@alt for blind users and the image itself for sighted users, since it
> passes "A")
>
> Thus, we have an image of text that is virtually invisible for anyone that
> has not a more-than-perfect vision (but passes A), and a link that is
> perfectly visible and passes level-A because the context is there.
>
> The described situation is more or less common, maybe the low contrast is
> not as critical, but I've seen many "iamge headings" with contrast ratios
> of 2.0:1 or similar, which are almost invisible for me.
>
> Therefore, I think that a technique similar to the "distinguisable links"
> should be created for images of text that have not enough contrast, and
> even a pair of failures of SC 1.4.1 due to "relying on color with an
> insufficient contrast ratio to convey information", or something like that.
>
> Kind regards,
> Ramón.
>
>
>