Replies: 21 comments 17 replies
-
|
the "text that is part of a picture that contains significant other visual content" part is meant to exclude "incidental" text - say, in a photograph of a person, somewhere in the background you can see a street sign with text, but it's not an essential part of what the image is trying to convey ... in that case, that text is not subject to the requirements. |
Beta Was this translation helpful? Give feedback.
-
|
This is how I understood the intent of the criteria, but the definition of 'image of text' suggests that including other significant elements might exempt a significant text from being considered an "image of text" |
Beta Was this translation helpful? Give feedback.
-
|
Correct. Image of text only applies to an image that only contains text. |
Beta Was this translation helpful? Give feedback.
-
|
Sorry Gregg, but that makes no sense. You saying that an image that contains text, as well as an icon, now doesn't count as "Image of Text" and doesn't need to follow, say, contrast guidelines? I don't believe that's the intent. An image of an infographic that includes text, arrows, box-outs ... you saying the text doesn't need to have sufficient contrast here either? |
Beta Was this translation helpful? Give feedback.
-
|
Inspired by this, working on a proposed erratum #5013 |
Beta Was this translation helpful? Give feedback.
-
|
I am sorry that it does not make sense to you. And no - I did not say that text in infographics should not have sufficient contrast. But I am saying that the current requirements do not require "text in an image" to meet contrast requirements. And the only place "images of text" is used normatively in WCAG in two places 1.4.5 and 1.4.3
And in the CONTRAST provision 1.4.3 - where it uses the same definition of "images of text" and has a specific exception for text in an image. Note that one of the exceptions in the CONTRAST requirement is
I would judge that infographics have significant other visual content -- or they would not be infographics. So while I agree that text in inforgraphics SHOULD have high contrast -- WCAG 2 does not appear to REQUIRE them to. AT least not in my reading. Best |
Beta Was this translation helpful? Give feedback.
-
|
we'll have to agree to disagree on those points then. because again, there have then clearly been diverging interpretations in the real world on when the contrast requirements do or don't apply. |
Beta Was this translation helpful? Give feedback.
-
|
as for the exception in the normative wording for 1.4.3 Contrast (Minimum) and 1.4.6 Contrast (Enhanced), it comes under the "Incidental" exception. The word "incidental" generally points to:
https://www.merriam-webster.com/dictionary/incidental so again, this would reinforce the idea that it's not every text that is present in an image with other visual content, but only when it is ... incidental |
Beta Was this translation helpful? Give feedback.
-
|
But still, the criteria only apply to text and images of text only
Here is my problem, we all understand that the exception only applies to incidental text in images of text, but the criteria still don't apply to "text in images" Either we should change the definition of images of text, or we should include text in images with the following note:
|
Beta Was this translation helpful? Give feedback.
-
|
I agree with Patrick that 1.4.3 contrast requirement does apply to text in images unless it is incidental. If it didn't apply to images of text unless there was no other image aspects in there then it would have said that - instead the incidental definition was crafted to clarify which situations of text didn't have to meet contrast requirements. This has been the long held stance of the working grouping - this is the first I am learning of Gregg's interpretation that only images containing text and not other content would be covered. I believe that most people are not failing incidental text when it has insufficient contrast - but text that is necessary to understand the content such as an info graphic, button with text and icon and that doesn't fall under some other exception. |
Beta Was this translation helpful? Give feedback.
-
|
@mraccess77 I agree that all text should have sufficient contrast. But the wording says that if there is any significant other content - then it is in the exception category. Could have been worded differently -- but it was not. Personally I think it is good to have all text have high contrast (actually I think all text should have variable contrast - to fit the user). Signing off on this. Was just passing on the original intent. |
Beta Was this translation helpful? Give feedback.
-
is there any explanation of why the group, at the time, left this gaping gap? |
Beta Was this translation helpful? Give feedback.
-
|
@patrickhlauke It didnt really leave a gap -- it just decided that it was impossible to find a clear objective line between something that had text and some graphics and something that had graphics and some text. And to make a requirement - it had to be objective. Again I can't wait until we have all of these be user choosable / settable in the browser instead of trying to require one value for all. ESPECIALLY since we have always had users with cognitive, language, and learning disabilities asking us for a requirement for low contrast as well..... |
Beta Was this translation helpful? Give feedback.
-
|
I don’t understand why there needs to be such a distinction. Surely, if an image contains important text, it must have sufficient contrast. Why does it need to be more complicated than that?
The same image might also have important non-text content, which would be subject to the lower contrast requirement, so different contrast requirements apply to different parts of the image.
If either the text or the non-text content is unimportant or incidental, the exemption would apply to that part of the content. Am I missing something?
[edited to remove the parts that came from this having been an email reply - the quoted original message, email signature, etc]
|
Beta Was this translation helpful? Give feedback.
-
|
A few points:
The definition: |
Beta Was this translation helpful? Give feedback.
-
|
I see two different use cases clashing here: A movie poster: It contains text like the movie title, but also the actors' names and other credits. Usually, the alternative text only covers the movie title because we consider it the only significant information (and only if this information is not already present on the page). In this case, the contrast analysis might only need to apply to that specific title. A commercial flyer published as a JPG: It might contain a store map (which qualifies it as having 'other significant content'). As it stands, 1.4.3 seems to suggest that because there is other significant content, the text doesn't need to meet contrast requirements. I think we would expect the text to say: 'when the meaning (the one conveyed in the text alternative) is carried by other significant content... |
Beta Was this translation helpful? Give feedback.
-
|
Much of this is good discussion for WCAG3 - and how we word that to avoid any gap. But for WCAG2 - we have to go with the normative wording (which does not include labels, headings etc.). So the word "incidental" is not normative. Only the sentence following it. Also - a accurately pointed out -- NOTES are not normative. So the Note that contains an exception is actually a mistake. If doing errata - that should be included as the end of the sentence fragment that makes up the definition. Unfortunately the term "images of text" was written to apply to situation where people used an image of text instead of formatting the text -- hence the way the provision and definition was written. It then got used in the contrast provision without changing it. But it would have been hard to figure out how to word it so that it would cover important text in an image from other text in an image. ALT text -- is actually a completely different topic - since contrast is for people with low vision to see the text - whereas ALT text would not be visible to them at all. What is needed in WCAG3 is some better language that can separate OBJECTIVELY incidental text from important text. AND can also handle the situation where the image is a picture -- and the author has no control over the contrast of text with its background. But I don't know what we can do for WCAG2 and I'm not sure I can see any clear way to handle it in WCAG3 yet either AGAIN - I say that what we need is smart browsers that can allow users to adjust the contrast of any text anywhere on any page including any images or photos - to match a user's needs for high medium or low contrast. |
Beta Was this translation helpful? Give feedback.
-
|
i mean the "objectively" part is fairly straightforward (though maybe not "automatable", so we'll have to wait a bit longer for the "AI browsers will just solve this" jetsons flying car future): if the text were to be removed altogether, would the image (or rather, the page/screen overall) still convey the same information/purpose? |
Beta Was this translation helpful? Give feedback.
-
However, we should also think about movie posters or any other artwork... Text could be 'non-incidental' without necessarily conveying relevant information, yet a web author might have no control over it... That being said, I agree with the consensus here: there's currently a loophole. |
Beta Was this translation helpful? Give feedback.
-
|
@patrickhlauke wrote:
I think the extent to which an image conveys the same meaning as a text is variable so not easy to determine objectively -- it is a sliding scale (from most conventional to arbitrary). Focusing on icons as simple cases, I just put these examples here to demonstrate what I mean, not as an argument over any particular one of them...
|
Beta Was this translation helpful? Give feedback.
-
But if something like a button is explicitly designed to have an icon and accompanying text, then it's not "incidental". They added the text to those buttons for a reason. They didn't add it as a visual flourish, or text that just happens to also be in the image. They made the image explicitly to provide both the text and the icon. Using common sense, this should not be difficult to discern from "it's an image that, as part of a photo, or some illustration work, or whatever, happens to have some text there that doesn't really convey anything relevant". |
Beta Was this translation helpful? Give feedback.
-
|
@patrickhlauke good point, so not incidental. The question remains if you'd pass a text label failing contrast if there is a conventional icon (think trash can) next to it which you'd pass on its own - i.e., when to consider stuff non-conforming alternate versions of conforming stuff. |
Beta Was this translation helpful? Give feedback.
-
I object to this as it is not what it says or what people do or what is useful. an inforgraphic, for example can include images of text that are required to understand the meaning of the infographic, the images of text that provide meaning must pass color contrast requirements. |
Beta Was this translation helpful? Give feedback.
-
|
Until the definition of 'images of text' is revised to cover a broader scope, these instances will remain misclassified under the category below, despite the obvious contradiction:
|
Beta Was this translation helpful? Give feedback.
-
|
I want to offer a common use case that may (or may not) have some relevance in this discussion and then ask a question. In PDF we commonly deal with (a) "designed" documents (that is, with fancy layouts), and (b) "scanned" documents (that is, page-images captured from scans). In both cases (but for very different reasons) it's quite typical to encounter "images with text" that express with color-contrast issues. PDF includes a facility that HTML doesn't; the concept of "ActualText", which is defined in PDF's specification as "Text that is an exact replacement for the content enclosed by the structure element and its children." Much like the more familiar (to HTML folks) Alt, an ActualText value may be established for any given image in a PDF. Today, ActualText is broadly (although not yet universally) understood by Readers / APIs / AT to provide in-line text, including at the level of a single character. Although ActualText solves the problem of "images with text" for text consumers (such as screen readers and TTS), I suspect that it's almost (but not entirely) orthogonal to the color-contrast issue, even if ActualText provides AT with machine-readable text in such cases. I'd be curious to read others' thoughts on the ActualText facility in the context of the color contrast question. In particular, can the presence of ActualText mitigate a color contrast problem from a WCAG point of view? |
Beta Was this translation helpful? Give feedback.
-
|
@DuffJohnson as you say, it's orthogonal to the discussion here I'd say
Unless there is a way for a user who struggles with low contrast text, but does NOT use AT such as a screen reader, to take advantage of this ActualText, it's just as useless to them in PDF as alt text is to them in HTML? |
Beta Was this translation helpful? Give feedback.
-
|
Yes... more-or-less as I thought. There is "a way" in that software can replace the rendered image with the ActualText (as happens, for example, with a copy/paste), but that's insufficient for this problem; fair enough. |
Beta Was this translation helpful? Give feedback.
-
|
as an additional point for discussion, see the recent changes/tweaks to Images of Text #4021 that were agreed by the working group . In particular, the additional example given for the image of a poster. this change addresses situations where text is part of a larger image that contains other (visual) content in the image. if that isn't supposed to count as subject to images of text, then this change would also need to be changed/backed out again. /cc @alastc @giacomo-petri |
Beta Was this translation helpful? Give feedback.
-
|
In actual practice as auditor, my experience is that it has not been to hard as to when to insist on 3:1 or conclude that 1.4.3 is not applicable. Is there reading of paragraphs or just words? Is the context something as complicated as a road map? No, that‘s not part of the SC, but it gives one insight as to when to insist on the stricter requirement. If a client balks at their low contrast wordy slides needing to meet 3:1, they don’t certainly don’t get the excuse of text being “part of a picture that contains significant other visual content." A good infographic like the DC Metro Map will provide as much contrast for all textual elements as possible. There will almost certainly still be some text that doesn’t meet 3:1. If I was asked to audit that DC Metro Map, I would judge 1.4.3 as not applicable, and 1.4.5 and 1.4.11 both as passing. |
Beta Was this translation helpful? Give feedback.
-
|
if you are leaning on 1.4.3, then it should be |
Beta Was this translation helpful? Give feedback.
-
Where I have had text contrast come up (for images of text), it’s typically in the context of PPT slides or portions of a video presentation. In those mediums, arguing over if its large text or not is so productive. Also, pointing out the 4.5:1 contrast requirement but then ”settling” for 3:1 is a win. Because 4.5:1 is so limiting, not settling for 3:1 might lead to an argument that 1.4.3 is not applicable. In the federal space, based on 2.0, we don’t have 1.4.11 as a backstop. |
Beta Was this translation helpful? Give feedback.
-
|
+1 to Bruce’s comments
It is always good practice to always have good contrast. But the question of whether or not it is required legally by any regulation that cites WCAG is the question. And there we need to rely not on what is best practice but what is actually the minimum required by WCAG. And for that we need to use the language.
I doubt that anyone is going to end up in a court of law over this one provision. Or it's interpretation.
So I would suggest that we not get too wound around the Axel and instead as part of an audit just say that it should have good contrast so that everybody including old people and those with color blindness (and old people with color, blindness and low vision) can clearly read whatever it is they are presenting.
… On Mar 23, 2026, at 11:02 AM, Bruce Bailey ***@***.***> wrote:
In actual practice as auditor, my experience is that it has not been to hard as to when to insist on 3:1 or conclude that 1.4.3 is not applicable. Is there reading of paragraphs or just words? Is the context something as complicated as a road map?
No, that‘s not part of the SC, but it gives one insight as to when to insist on the stricter requirement. If a client balks at their low contrast wordy slides needing to meet 3:1, they don’t certainly don’t get the excuse of text being “part of a picture that contains significant other visual content."
A good infographic <https://en.wikipedia.org/wiki/Infographic> like the DC Metro Map <https://en.wikipedia.org/wiki/Infographic#/media/File:Washington_Metro_diagram_sb.svg> will provide as much contrast for all textual elements as possible. There will almost certainly still be some text that doesn’t meet 3:1. If I was asked to audit that DC Metro Map, I would judge 1.4.3 as not applicable, and 1.4.5 and 1.4.11 both as passing.
—
Reply to this email directly, view it on GitHub <#5012?email_source=notifications&email_token=ACNGDXXR6DTCQO4BZN5YIGD4SF3ZRA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNRSHAYDEMBXUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-16280207>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACNGDXR3QF3B5XNKBXW7AGD4SF3ZRAVCNFSM6AAAAACW2GPRRSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMMRYGAZDANY>.
You are receiving this because you were mentioned.
|
Beta Was this translation helpful? Give feedback.
-
|
The term incidental is misleading if we don't mean incidental then that word should not be used. If we do mean incidental then that should be clear that it only applies when incidental - which is how many of us understood it. What's frustrating to me is that this could have been addressed in 2.1 and 2.2 - no one spoke up until 18 years later to explain that this really isn't about excluding incidental text but instead excluding all images of text that have non-text content. Had this position been described earlier we would certainly have addressed it in 2.1 in 2018. Regarding whether we should just report the finding - in my experience despite our efforts as a community to educate people - most organizations specifically ask to not have advisory issues reported - and most organizations focus only on hard failures of WCAG. While we all know it should be the floor and not the ceiling most organizations will only address something if it's a WCAG failure and they will very much come back to this thread to argue why they don't need to do it. |
Beta Was this translation helpful? Give feedback.
-
|
Agree. we should not have used the word in the title when it is not actually included in the Normative part. It makes it sound like "incidental" is part of the judgement - but it is not actually in the normative. Need to fix in w3 |
Beta Was this translation helpful? Give feedback.
-
|
On a related subject: Success Criterion 1.1.1 is titled "Non-text Content" (Non-text Alternative), while 1.4.11 refers to "Non-text Contrast". In your opinion, is "non-text" being used in the same way in both contexts? IMHO, 1.4.11 should have been "non-textual content" contrast, but am I missing a nuance here or is it just a simplification? |
Beta Was this translation helpful? Give feedback.
-
that, in my mind |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello,
I have always considered "standard" text in images as being covered by SC 1.4.5, but I think I might be wrong...
The definition of "images of text" is:
This seems to exclude, for instance, a flyer also containing other significant visual content like a QR code, photographs, or graphical objects covered by SC 1.4.11.
Should we consider the text (parts of graphics) in such examples to be only covered by SC 1.4.11?
Beta Was this translation helpful? Give feedback.
All reactions