RE: Study: The effect of serifs and stroke contrast on low vision reading

Dear Andy

Thanks for the additional detail, very interesting. Only one follow-up point to your statement

  *   I've never once in my life, and this includes working in the print industry when I was young and in traditional print typography and offset printing, never once have I heard of "too high contrast" printing black ink on white paper.
I certainly have heard of 'too high contrast' printing black ink on white paper. Arnold Wilkins has done lots of research in this area, mostly relating to how children in education perceive printed text. He found that adding a coloured overlay on top of the printed text reduces the white of the background, thus reducing the contrast difference between foreground and background. Different children found different coloured overlays to be the most effective, but lots of them found the experience completely transformative, from being unable to read books to loving reading them. If you search for visual stress and Arnold Wilkins you will find lots of published material on this topic.

So, is not just an issue of screens, visual stress does also occur for printed materials. I would however agree that the problem may be more significant for emissive displays. In solving the problem for emissive displays, I'm currently advocating the following solutions, in this order:

  1.  switch to a dark mode colour scheme
  2.  use black text on a grey background
  3.  use grey text on a white background

As a slight tangent, I did another project many years ago looking at projector screens in UK schools, and most schools project directly onto normal whiteboards. In an ideal world this should never happen, as whiteboards are awful for projecting onto, and a pulldown screen would be a much better job of displaying the content. However most classrooms make do with what they've got, which is a projector and a normal whiteboard. Whiteboards are extremely reflective, and slide decks with pure white backgrounds cause a huge white spot reflection to appear in some part of the projected content, which is extremely unpleasant to look at. Switching the slide deck to dark mode solves this problem almost instantaneously!

Best wishes

Sam Waller (he, him)
University of Cambridge, Engineering Design Centre
01223 332826

From: Andrew Somers <andy@generaltitles.com>
Sent: 03 January 2023 16:49
To: Jon Avila <jon.avila@levelaccess.com>; Sam Waller <sdw32@cam.ac.uk>
Cc: public-low-vision-a11y-tf <public-low-vision-a11y-tf@w3.org>
Subject: Re: Study: The effect of serifs and stroke contrast on low vision reading

Hi Jon, and Sam,

One of the things I'm looking at, is statements regarding "high contrast" that are not necessarily related in terms of environmental adaptation considerations.

What I mean here is, what is often being referred to as "high contrast causing fatigue" is not the contrast per se, but rather the high luminance levels relative to the adaptation state.

High contrast with high luminance levels, but associated with a low adaptation state, is certainly fatiguing, and I don't believe that relates to any particular impairment, I believe that applies to everyone, at least to some degree as a matter of a spectrum, and we could say can be related to disability glare.

Something that I've been working on, the Paper Reading Experience Project, is concerned or involved with defining these properties in a quantitative and qualitative way. Ultimately the goal is some general design guidelines for displaying readable content on a self illuminated device that reads like paper.



Anecdotal:

Decades ago before my personal vision issues, I much preferred reading on paper than reading on a computer screen.

My current vision makes reading on paper very difficult simply because standard print is much too small even with glasses and a magnifying glass it's challenging. But I have challenges reading on a computer screen even with text zoomed 400%, due to luminance versus adaptation state, and for too many websites using low contrast text, etc. etc.

A point I'm getting out here though is, in traditional print we always used black ink on light or white paper. Literally, maximum contrast has always been a desirable standard in the print industry.

Something that I think has been happening over the last 15 years, as websites started using lower contrast text (after WCAG 2.0 said that 4.5:1 was "OK," and sites subsequently moved from black to gray), in order to read the web now, people have been making their monitors brighter. The brighter monitors cause higher luminance relative to the adaptation state, therefore causing greater fatigue.

In other words, it's something of an uroboros, a dragon eating its tail.

There are people out there pushing design guidance, asserting the way to reduce contrast is to make text lighter gray, but not doing anything to modulate the excessively white background. My response is that such guidance is exactly wrong (I discussed this in "Please Stop Using Gray Text" an article that surprisingly went viral a few months ago).


Related to this, here are a couple of points, some of these are not always well considered when they should be.


  *   Screen resolution in association with the rasterizing or rendering technology directly affects the actual rendered contrast due to antialiasing effects.

     *   On a standard resolution 72 or 96 ppi display, normal weight body text, i.e. 16px, is substantially affected by antialiasing effects, causing the major strokes in each rendered glyph to have a reduced luminance distance to the background.
     *   On a high resolution retina display, major strokes of a normal weight font at 16px, are more likely to be rendered with at least one or two pixels at the actual assigned CSS color.

  *   Regardless, for a block of body text as above, with a white background, the difference between #333 and #000 on a white FFF background is negligible, and in brighter environments, often less than JND, essentially invisible. YMMV. (I'm going to use 333 in the examples below, but depending on the monitor adjustment, #222 could be the key number).

     *   In a typical office background of about 350 lux, screen flare accompanied with ocular glare from a white background and related factors, pretty much wipes out any difference between #333 and #000 for small thin blocks of body text.
     *   Find here that there is no functional difference, and absolutely no advantage to using 333 versus 000 for body text with a white background.

  *   On the other hand, for text at either #333 or #000, changing an #fff background to #eee or #ddd has a remarkably significant effect on the reduction of eyestrain from excessive luminance.

     *   Due to contrast constancy, even with 16px body text, as we get the contrast above around Lc 85, we start having diminishing returns in terms of increasing the "contrast" as it begins to force increasing the overall luminance.

  *   Make no mistake, the background behind the text and the larger proximal background of the total screen do have a significant effect on the overall adaptation state of the eye.

     *   But keep in mind, that in paper reading, the reflectivity of the paper is at most 90% of the ambient (for a glossy white magazine).
     *   Reflectivity of a typical book is closer to 80 to 85%, and is highly diffuse. A newspaper is closer to 75% or less.

  *   Depending on the adjustment of the monitor, the peak white could be be somewhere between 102% to 120% or more of the reference adaptation state, relative to what is defined as the reference environment for reflected colors for Munsell which is what CIELAB L* is based on.

     *   In other words, assuming an environment as defined by Munsell when we're reading a typical paper book, we're looking at page white that is essentially 85% and never more than 90%, and very diffuse.
     *   When we're looking at a self-illuminated monitor with the background of white FFF, we're only looking at 102% to 120%+ and not at all diffuse but beaming straight into our eyes, and often with polarized light.

  *   I've never once in my life, and this includes working in the print industry when I was young and in traditional print typography and offset printing, never once have I heard of "too high contrast" printing black ink on white paper.

     *   But it is an issue with computer displays because they emit their own light as opposed to diffusely reflecting ambient light, and a computer display is capable of emitting substantially more light than what would nominally be diffusely reflected by a paper page.
     *   Because the self illuminated monitor contributes to the adaptation level, it makes this more complicated to calculate.

        *   In other words if we just reduce the background of the webpage from FFF to DDD, then DDD ends up becoming the significant peak of the adaptation state, and over a period of a couple minutes becomes indistinguishable from white, because we have adapted to it... until we switch back to FFF in which case FFF suddenly seems blazing bright


With these factors in mind, the idea behind The Paper Reading Experience is to use white FFF in the periphery to anchor the adaptation state, so that the background immediately behind the text can remain closer to an optimum 85 to 90% relative luminance.

There are some examples of this concept on the following experimental page link:


https://www.myndex.com/WEB/LuminanceContrastLIGHTMODE<https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.myndex.com%2FWEB%2FLuminanceContrastLIGHTMODE&data=05%7C01%7Csdw32%40universityofcambridgecloud.onmicrosoft.com%7Cd2d7666b3ab043ab1ba408daedaa64d5%7C49a50445bdfa4b79ade3547b4f3986e9%7C1%7C0%7C638083613338846388%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=dOaL6aNCnOPBR7RwsqSJ%2FmE2ZQNWngNaN%2FVCggQE7h0%3D&reserved=0>

Note that the first 10 or so paragraphs in this article cycle through three colors and three weights, which can be viewed using the inspector and the browser's developer tools.

The cycled colors starting with the second paragraph in the actual article, are #000 #333 #007 and repeat for each of the the cycled weights 500 400 300.




Thank you for reading,


Andy






Andrew Somers
Color Scientist @
Myndex Research
W3C Invited Expert
WCAG 3.0 Co-Author






On Jan 2, 2023, at 3:22 PM, Jon Avila <jon.avila@levelaccess.com<mailto:jon.avila@levelaccess.com>> wrote:

Hi Andrew, I understand what is being said about low contrast stroke.

When I speak to people who prefer low contrast (lightness) text - I believe it's not about recognition of the text but about the impact that high contrast has on the visual system.  Some people report that high contrast patterns can cause migraine like symptoms.  So it's more about the impact which prevents them from staring at high contrast lightness text.

Jonathan

From: Andrew Somers <andy@generaltitles.com<mailto:andy@generaltitles.com>>
Sent: Monday, January 2, 2023 3:04 PM
To: Laura Carlson <laura.lee.carlson@gmail.com<mailto:laura.lee.carlson@gmail.com>>
Cc: public-low-vision-a11y-tf <public-low-vision-a11y-tf@w3.org<mailto:public-low-vision-a11y-tf@w3.org>>
Subject: Re: Study: The effect of serifs and stroke contrast on low vision reading

And to follow up my previous post I wonder if misunderstandings relating to the meaning of "font-stroke-contrast" versus actual "visual contrast" is what led to this weird myth that seems to persist on the internet today, claiming that "some people need low contrast text".

???

I have been searching for credible research that might affirm the assertion "some people need lower contrast" but I'm not finding much relevant ... And I am increasingly thinking, that some may have read "low contrast font" in some paper but are misinterpreting it to mean low contrast colors, when it actually means having a font with a more uniform stroke width.





On Jan 2, 2023, at 11:40 AM, Andrew Somers <andy@generaltitles.com<mailto:andy@generaltitles.com>> wrote:

Thank you Laura, that's a very interesting study, and directly echoes my personal viewpoint on this particular subject.

I want to make a note here for anybody that reads this study:

The study uses the term "stroke contrast". It's very important to recognize that in the context that they are using it, stroke-contrast relates to the variation in stroke within a particular glyph.

And the effect is exactly opposite of the actual visual contrast.

In other words in the context of font contrast as in the stroke-contrast within a glyph, e.g. Times New Roman has very high stroke contrast.

But the visual contrast of Times is lower than that of Helvetica. Helvetica, having a uniform stroke width therefore has a low stroke contrast but (perhaps counterintuitively) that results with its visual contrast as higher.

Here's an example:

<Screen Shot 2023-01-02 at 11.35.14 AM.png>





On Dec 27, 2022, at 8:17 AM, Laura Carlson <laura.lee.carlson@gmail.com<mailto:laura.lee.carlson@gmail.com>> wrote:

Fyi:
https://www.sciencedirect.com/science/article/pii/S0001691822003250<https://eur03.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.sciencedirect.com%2Fscience%2Farticle%2Fpii%2FS0001691822003250&data=05%7C01%7Csdw32%40universityofcambridgecloud.onmicrosoft.com%7Cd2d7666b3ab043ab1ba408daedaa64d5%7C49a50445bdfa4b79ade3547b4f3986e9%7C1%7C0%7C638083613338846388%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=r95dX27rZ%2BmURjdOLN53PwhozqzNADwd2y%2BkDw90Iug%3D&reserved=0>

Kind Regards,
Laura

--
Laura L. Carlson