Re: Colour contrast algorithm when black is used

Hi there,

> Black against #777777 passes, and black against dark orange passes, but to the eye, white #FFFFFF which fails, seems much more readable against this grey  or against this orange.

I also feel like this is a risky thing to do. As someone who is colorblind, I would argue the exact opposite(1). Black text on a mid grey or dark orange is very readable and stands out a lot more to me. Arguably, I've learned a long time ago not to trust my own color perception, but I suspect we all experience color perceptions slightly differently.

(1) This is the representation of the experience of one user who is color blind, not all users who are. 


Denis Boudreau,
Principal accessibility consultant & trainer
Deque Systems, Inc.
Cell: +1-514-730-9168
Email: []

Keep in touch: @dboudreau []

On 2017-08-21 4:27:28 AM, Alastair Campbell <> wrote:
Hi David,
I think we would be in very dangerous territory if we try and patch the algorithm without a lot more data.
Just anecdotally from testing, I’ve seen at least two groups (which probably have sub-groups) who’s perception of contrast differs from mine: low-vision and older-people.
Strong colours to my vision were muted or invisible to them, and in one-case the other way around. For example, a colleague with red/green colour blindness can read dark-blue on black easily where I can barely tell there is text there.
As people (Jared, Amelia I think?) mentioned, there are differences in perception due to hue, but unless we’ve got research around the whole colour gamut for lots of different ‘vision’ types, I don’t think we should try patching the algorithm.
The advantage of the hue-less algorithm is that it ‘works’ regardless of hue perception. Not perfectly, but to some degree that makes an improvement in general across groups.
From: David MacDonald
Could be...  
for a 2.1 we're probably not going to solve it all, unless we're super lucky... I'm thinking about a minor tweak from what we know now that would have broad appeal and plug significant holes.
So if there's a simple amendment in a colour space, that would be awesome... I'll be interested in Jared's suggestions.

David MacDonald
CanAdapt Solutions Inc.
Tel:  613.235.4902
[] []
GitHub [] []
  Adapting the web to all users
            Including those with disabilities
If you are not the intended recipient, please review our privacy policy []
On Sun, Aug 20, 2017 at 1:10 PM, Patrick H. Lauke < []> wrote:
On 20/08/2017 11:41, David MacDonald wrote:
There recently was a Twitter discussion about colour contrast combinations. []

There seems to be consensus from stakeholders that when black #000000 is contrasted against some colours, the 4.5 threshold is met even when it's hard to see. I've noticed this for years, and it seems others have too, but the Twitter discussion got me thinking about it again.

Black against #777777 passes, and black against dark orange passes, but to the eye, white #FFFFFF which fails, seems much more readable against this grey  or against this orange.

There is something about black #000000 which needs a little tweak in the algorithm.

Perhaps when black is against colour spectrum X, it requires a higher contrast minimum?

Are you sure it's literally *just* full #000000 black? What about #000001? My point being that it's perhaps not just about that one particular color, but more about a particular part of the spectrum / the color space used?

Patrick H. Lauke [] | [] [] | []
twitter: @patrick_h_lauke | skype: patrick_h_lauke

Received on Monday, 21 August 2017 12:58:39 UTC