IRT • Inclusive Reading Technologies
Replies: 7 comments 8 replies
-
|
Is there an a more detailed explanation of non-polar mode?
Is there a minimum font size for Lc < 90 and >= 75?
I think using the x-height of a font might be a barrier to simplicity and ease of use. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @lloydk Thank you for your comments:
It's pre-release, but basically, it is essentially the result that you get when you always calculate the text as darker than the background (in other words, the darkest color always goes into the text input regardless of how that color is being used). It may be such that it checks both polarities and then always reports the lowest contrast. The idea behind non-polar mode is to simplify.
We haven't quite settled, but are leaning toward 18px. We almost could say that 16px is possible, except when you look at a font such as RALEWAY, its 400 weight is equivalent of 300 weight vs the reference font. And since Bronze Simple is intended to be as simple as WCAG2, we are struggling with not using a reference font for a comparison, yet still providing meaningful results. And actually I'd like to hear commentary regarding using minimum font sizes for Lc 75+:
Choice one is what I would prefer.
Thank you for pointing that out, that was copy-pasta from the master use-case description, edited now for simplicity. |
Beta Was this translation helpful? Give feedback.
-
|
If Silver, Gold, etc. keep the concept of polarity then I think overall the spec will be more confusing than if polarity was either in all conformance levels or ditched entirely. I'd prefer to see it kept in all levels without mentioning the word polarity which is a technical term that some people might not understand. I think that the negative values returned for WoB could be considered confusing so returning a contrast value >= 0 for BoW and WoB might help to simplify the concepts surrounding the APCA algorithm. Personally, I've never had a need to know what the polarity of the contrast value is and the fact that the value might be negative is mostly an inconvenience. |
Beta Was this translation helpful? Give feedback.
-
Hi @lloydk Thank you for commenting, non-polar mode it's literally brand new and experimental, it was an idea we started playing with very recently, looking for ways to find the absolute simplest reduction to practice. That said, In order to get the benefits of design flexibility and best readability for silver & gold levels, polarity definitely needs to be a part of that. Instead of the term polarity is there another term that you would suggest?
If you have thoughts here we'd love to hear them.
The substantial majority of users seem to prefer showing polarity as positive or negative, but I have noticed some edge cases with people who have never worked with APCA before being initially confused, which is why I did that poll on Twitter. But this doesn't happen very often. And we've improved and simplified the documentation. quick question for youAnd that brings up a point I'm very curious about, what if any of the documentation have you happened to read? And which tool or tools are you using?
I don't expect designers Will necessarily need to know the polarity. Coming from a design point of view, particularly if you're using a design tool like Leonardo, as long as you know which color is designated text and which color is designated background. As I indicated on that thread in Twitter, the reason for using a negative number is partly because it's actually natural to how the math naturally functions. But in the written specification I also point out that what's important is that the text color and the background color are properly identified. One way to figure out which color is which, if the number is negative then you know that the background is to be darker than the foreground, and vice versa. TestersBut while a designer using a design tool may not necessarily need to know what the polarity actually is, somebody who is doing testing of a site to see if it is conformant is most likely going to want to see polarity as a function of testing the site. |
Beta Was this translation helpful? Give feedback.
-
|
I think by trying to simplify bronze level by introducing non-polar mode you're going to end up introducing more complexity in the the entire accessibility ecosystem (specs, documentation, tooling, etc.). Not a big fan of the light mode/ dark mode terms for polarity. Those terms are commonly used to describe css color schemes. Just google
I've read all of your documentation and articles many times over. I've tested hundreds of color combinations using your APCA contrast tool. Developed a bunch of small sites trying to conform to the suggested APCA contrast ranges and font sizes. Implemented the APCA contrast function in GLSL and developed some APCA/WCAG comparison tools which may or may not get released publicly. |
Beta Was this translation helpful? Give feedback.
-
Hi @lloydk Thank you for that feedback, this is the kind of feedback I wish we had more of… As I said we just brought up the idea of a non-polar mode, and we haven't even put it into any tools or code yet for testing and experimentation. The reason: we are faced with a number of challenges, those challenges include integrating methods into existing testing tooling, and finding entry points for sites that are currently using WCAG 2 to make it easy to shift over. Also there are there legitimate cases for contrast between two elements that are not text or at high spatial frequencies. In these cases, a non-polar approach may provide certain benefits, and that also is part of why we were exploring some of these options.
Excellent, thank you for letting me know. Regarding your comments here or on Twitter for polarity indication, do you find anything that you do is being blocked or difficult because of the current polarity indicator? (i.e. minus for negative polarity), or is it just that you feel it's purpose is not clear? Regarding the term "polarity", you mentioned concerns about this as a term, is there another term that you think would be more appropriate? Can you share any links to any of the sites that you put up? |
Beta Was this translation helpful? Give feedback.
-
I would say that it's clear (at least to me) what information the polarity indicator provides but as far as I can tell there's currently little if any use for it. The guidelines produced so far (the contrast ranges for various use cases) don't mention it so as I said on Twitter I'm not sure what I'm supposed to do with the polarity indicator. I say that as a user of your online contrast calculator, a developer of tools and a tester. In all of those scenarios I already know which colors are foreground and background and don't really care if the foreground is lighter or darker than the background. You've alluded to upcoming developments that make the polarity indicator more relevant so maybe when that work comes I'll have to take the indicator into account when checking against contrast ranges for various use cases.
My only concern with the term "polarity" is that in a spec with diverse audience it might be to technical but I don't have any other suggestions. The main thought I had was the opportunity to simplify things by removing the polarity indicator entirely but it sounds like that won't be possible due to the upcoming changes. |
Beta Was this translation helpful? Give feedback.
-
Ah, I see, this is why I asked which documentation you've read thus far, as this sounds like a documentation issue... Documentation for this project is still a work in progress like several other things, so that's something that needs to be discussed I think in a little bit more detail. Thank you |
Beta Was this translation helpful? Give feedback.
-
UpdateJust updated the levels as shown below, above however, the descriptions were made more terse. Note this is not "WCAG 2 backwards mode". APCA-RC Bronze Simple Level Conformance
|
Beta Was this translation helpful? Give feedback.
-
|
Are the font sizes really What does the the |
Beta Was this translation helpful? Give feedback.
-
|
Hi @lloydk Yes, >= in theory per the reference font—I am still taking feedback regarding reference fonts and font matching for Bronze level. The ± means the Lc value for either positive or negative mode. There are cases where there will be different threshold for positive or negative polarity for certain features. And incidentally, as I think you had asked regarding importance of marking polarity: we discovered it applies directly to maximum contrast. Lc -90 is maximum contrast, and because it is negative, it applies only to dark mode polarities. While there is no maximum contrast for light mode, there is a maximum suggested brightness for backgrounds in light mode which will be discussed in the not too distant future. |
Beta Was this translation helpful? Give feedback.
-
Does the dark mode maximum contrast apply to all use cases or only to selected use cases like body text? There's a fair number of colors that exceed Lc -90 with a white foreground color. I can see darker buttons with a white foreground color failing because they're > Lc -90 being an issue for some people. I just checked a few sites that I worked on and I have some buttons that are > Lc -90. I also have a lot of dark mode footers with 14px text that are > Lc -90. Lowering the contrast of one of the footers from Lc -101 to Lc 90 does in fact make it more readable. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @lloydk Right now the maximum of Lc -90 is intended for all text and semantic non-text, and possibly for all elements in dark mode. However, we want to conduct a larger study before saying this as an absolute. What I imagine at some point, is something like:
Again to be clear, these are some initial values that are under study. |
Beta Was this translation helpful? Give feedback.
-
|
Here's a codepen that shows more more less what the footer looked like. The footer has poor line height and centered text which contributes to the poor readability. I added an example at 14px with Lc -90 and an example at 20px with a more reasonable line height. I've also added some CSS variables for Lc -85, -80, -75. I find the 20px example at Lc -80 the most pleasant to read. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @lloydk As the max contrast is a theory issue, I've started a new discussion at the APCA forum: It would be cool if you posted your last bit with the codepen over there... |
Beta Was this translation helpful? Give feedback.
-
The very early drafts involving SAPC and later APCA involved the use of a font table that some found daunting. In an effort to bring perceptual uniformity to a wider group of users, we've worked to find the bare essentials, the absolute minimums that would still be useful as a most-basic and easiest-to-implement guideline.
The Irreducibly Short Bronze Simple Level
This is intended to be visible on or near a tool's UI as a quick reference.
Lc 15: Thick Non-Text
Lc 30: Sub-fluent Text
Lc 45: Large Text > 36px
Lc 60: Fluent Text > 24px
Lc 75: Fluent / Body Text > 18px
Lc 90: Preferred Body Text > 16px
(Sizes per a 0.52 x-height ratio)
The Slightly Embellished Bronze Simple Level
(For tooltips or when there is more space available)
Lc 15: Large / Thick Non-Text (no less than 5px thick)
Lc 30: Sub-fluent Text min / semantic non-text min
Lc 45: Large Text > 36px normal, 24px bold
Lc 60: Fluent Text > 24px normal, 16px bold
Lc 75: Fluent / Body Text > 18px normal
Lc 90: Preferred Body Text > 16px
Font sizes shown are for the reference fonts Arial or Helvetica.
Other fonts should be compared and adjusted for equivalence.
BRONZE SIMPLE LEVEL IN DETAIL
Edited Aug 20 2023 to indicate more font sizes and weights at each threshold.
Please comment if you feel this over complicates.
Bronze Simple Mode
Bronze Use Case Conformance Values for Testing
Lc values listed using an approved APCA compliant algorithm
APCA-RC Bronze Simple Level Conformance
¹ Small Fluent Text means content that is not otherwise body text
² Recommended minimum for Lc90, not mandated
The APCA tool will soon output a line weight value in addition to Lc value for non text and dataviz.
DataViz A
Non-Text & DataViz C
Complex Data Display
Pie Pieces, Bars, Maps
Chart Grids, Lead lines
Development
Understanding Bronze Simple Mode:
ARC Use Case Definitions for Fluently Readable Text
FLUENT
Note:
¹ Partial lines do not count toward the body-text block definition, so 2 ½ lines is not necessarily body text, unless it is a 2½ line paragraph as part of a larger block or column. Three full lines generally would be the bright line for body text.
² A full line is defined as a line of text at the average line width for the given text block, usuallhy no less than 35 characters, and preferably less than 60 characters per line average.
Beta Was this translation helpful? Give feedback.
All reactions