Comments on: Testing Color Contrast for WCAG 1 & 2

By: Joe Dolson

Joe Dolson — Fri, 03 Oct 2008 14:02:03 +0000

As mysterious as it is, it’s nice to know that I don’t have to recode anything! Glad you “solved” your problem.

(Don’t you just love it when you “solve” problems without having the faintest idea what you might have changed?)

By: Christophe Strobbe

Christophe Strobbe — Fri, 03 Oct 2008 00:49:17 +0000

Hi Joe,
I redid my spreadsheet with more intermediate columns (i.e. additional columns for the sRGB values) and the discrepancies are now gone. I still don’t know what was wrong with the original version. At least you don’t need to worry about your PHP function.

By: Joe Dolson

Joe Dolson — Thu, 02 Oct 2008 16:27:03 +0000

The font size isn’t part of the contrast comparison; it’s simply that large font sizes have a different boundary measurement. So although a font size of 18pts will pass where 12pts won’t, it’s not because the algorithm is different; just because the judgement is different.

It would be interesting to experiment with incorporating the font-size directly into the algorithm, and if the guidelines specified a sliding scale that would probably be pretty reasonable; but with the guidelines having only a couple of markers, it doesn’t really make any sense to incorporate that element into the algorithm itself.

By: Mike Cherim

Mike Cherim — Thu, 02 Oct 2008 16:18:32 +0000

I’m curious, when concerning WCAG2, the font size affects the requirements (bigger text, less contrast allowed). Is this considered? Is this possibly affecting the outcome?

By: Joe Dolson

Joe Dolson — Thu, 02 Oct 2008 15:37:02 +0000

Interesting…it’s certainly enough of a difference that you can be reasonably sure it’s not a rounding error!

Without seeing your spreadsheet, obviously I can’t make any practical comparisons between the two formulas - but something is clearly fishy!

By: Christophe Strobbe

Christophe Strobbe — Thu, 02 Oct 2008 13:24:09 +0000

Hi Joe,
Thanks for posting the function. My implementation is in an ODF spreadsheet. I can’t see anything wrong in your function. And after several rechecks I can’t see anything wrong in my spreadsheet. Go figure!

Below are a few colour combinations that I tested yesterday:
#F0FFD2 on #006400:
7.06:1 (Spectrum Tester),
7.06:1 (Color Contrast Comparison),
6.8 (my spreadsheet)
#FFFFD2 on #006400:
7.26:1 (Spectrum Tester),
7.26:1 (Color Contrast Comparison),
6.98 (my spreadsheet)
#EEEECC on #004B2D:
8.68:1(Color Contrast Comparison),
8.59 (my spreadsheet)

I discovered this while working on a WCAG 2.0 implementation. The first two combinations would meet WCAG 2 Level AAA according to your function, but would fail according to mine. Hmm.

By: Joe Dolson

Joe Dolson — Wed, 01 Oct 2008 19:32:15 +0000

Thanks for comment, Christophe! The actual function I’m using to test relative luminance is this:


function luminosity($r,$r2,$g,$g2,$b,$b2) {
	$RsRGB = $r/255;
	$GsRGB = $g/255;
	$BsRGB = $b/255;
	$R = ($RsRGB < = 0.03928) ? $RsRGB/12.92 : pow(($RsRGB+0.055)/1.055, 2.4);
	$G = ($GsRGB <= 0.03928) ? $GsRGB/12.92 : pow(($GsRGB+0.055)/1.055, 2.4);
	$B = ($BsRGB <= 0.03928) ? $BsRGB/12.92 : pow(($BsRGB+0.055)/1.055, 2.4);

	$RsRGB2 = $r2/255;
	$GsRGB2 = $g2/255;
	$BsRGB2 = $b2/255;
	$R2 = ($RsRGB2 <= 0.03928) ? $RsRGB2/12.92 : pow(($RsRGB2+0.055)/1.055, 2.4);
	$G2 = ($GsRGB2 <= 0.03928) ? $GsRGB2/12.92 : pow(($GsRGB2+0.055)/1.055, 2.4);
	$B2 = ($BsRGB2 <= 0.03928) ? $BsRGB2/12.92 : pow(($BsRGB2+0.055)/1.055, 2.4);

	if ($r+$g+$b <= $r2+$g2+$b2) {
	$l2 = (.2126 * $R + 0.7152 * $G + 0.0722 * $B);
	$l1 = (.2126 * $R2 + 0.7152 * $G2 + 0.0722 * $B2);
	} else {
	$l1 = (.2126 * $R + 0.7152 * $G + 0.0722 * $B);
	$l2 = (.2126 * $R2 + 0.7152 * $G2 + 0.0722 * $B2);
	}
	$luminosity = round(($l1 + 0.05)/($l2 + 0.05),2);
	return $luminosity;
}

So, yes, I am actually applying that adjustment. The information posted in the article is just the definition of the ratio, not the actual formula used to derive the ratio — it’s simplified. I certainly won’t claim that there can’t be a bug in that function - and if you see one, by all means let me know!

Can you give me an example of the difference you’re getting?

By: Christophe Strobbe

Christophe Strobbe — Wed, 01 Oct 2008 10:07:28 +0000

My own calculations of colour contrast with the relative luminance formula from WCAG 2 generate slightly different results from what I get with the colour spectrum tester. In the last section of your article (“Color Comparison Formulas”) you list the formulas you use, but the section on the WCAG 2 formula does not mention how R, G and B are defined. WCAG 2’s relative luminance definition also says: “if RsRGB <= 0.03928 then R = RsRGB/12.92 else R = ((RsRGB+0.055)/1.055) ^ 2.4”. You don’t mention this. Does that mean that you don’t use this in your tool?

By: Joe Dolson

Joe Dolson — Mon, 18 Aug 2008 20:54:44 +0000

Thanks! Glad it helped you out, Gustavo.

By: Gustavo Fischer

Gustavo Fischer — Mon, 18 Aug 2008 13:19:09 +0000

Thank you, the Color Contrast Spectrum Tester was exactly the tool I was looking for.