Color accessibility: contrast and color blindness, explained simply
Updated on · Reading time: 6 minutes · By Gabriel Maglia
A design can look beautiful and still be unreadable for part of your
audience. Color accessibility comes down to two things that are easy to
check: enough contrast so text can be read, and not
relying on color alone to convey information. Both can be
verified in a couple of minutes.
The WCAG contrast check: you combine text and background and it gives you the ratio with the AA/AAA badges.
1. Contrast: so the text can be read
Contrast measures how different the text is from its background. It's expressed as a ratio, from 1:1 (invisible) to 21:1 (black on white). The WCAG accessibility guidelines set minimums:
Normal text: at least 4.5:1 (AA level), or 7:1 for AAA (stricter).
Large text (headings): 3:1 (AA) is enough.
The typical mistake is light-gray text on a white background: it looks
"elegant" but many people can't read it. Before calling a combination good,
measure the ratio with a
contrast checker: it
tells you if it passes AA/AAA for normal and large text.
2. Color blindness: don't rely on color alone
About 8% of men have some form of color blindness (and a smaller percentage of women). The most common cases confuse red and green. That breaks patterns we take for granted:
A chart that distinguishes categories only by red/green colors.
A form where an error is marked by "turning the field red".
A status light (green/yellow/red) with no text or icons.
The golden rule: color should never be the only signal. Always add something more —an icon, a word, a pattern, a different brightness— so the information is understood even if the color isn't perceived.
How to check your design in 2 steps
Contrast: measure the text against its background with the checker; it should pass AA at minimum.
Color blindness:simulate how your design looks with protanopia, deuteranopia, tritanopia and in grayscale. If something stops making sense, that information depended on color alone.
Quick best practices
Body text always with high contrast; reserve light grays for secondary details, not for reading.
States (success/error) with icon + text, not just color.
In charts, combine color with labels, patterns or shapes.
Try your palette in grayscale: if everything is distinguishable, you're on the right track.
All in your browser
Both the contrast checker and the color blindness simulator run locally: you upload an image or pick colors and everything is processed on your device, without uploading anything.
Frequently asked questions
Does accessibility affect SEO or just users?
It helps both: it improves the experience for everyone and is a valued best practice, but the main reason is that more people can use what you make.
We tested it: the contrast checker against my own WCAG arithmetic
Image: I didn't use any image: the tests are on loose colors.
Tool: Color Studio → Palette → Contrast (WCAG).
What I did:
In Color Studio I entered five color pairs in “Text” and “Background” with the color picker (without loading any image).
I noted the ratio and the four badges (AA normal, AA large, AAA normal, AAA large) that it shows.
I calculated each ratio separately with the WCAG 2.x relative luminance formula and compared them.
What I measured:
#1f2937 on #d6f64b: 11.98:1 → ✓ AA · normal, ✓ AA · large, ✓ AAA · normal, ✓ AAA · large.
#777777 on #ffffff: 4.48:1 → ✕ AA · normal, ✓ AA · large, ✕ AAA · normal, ✕ AAA · large.
#767676 on #ffffff: 4.54:1 → ✓ AA · normal, ✓ AA · large, ✕ AAA · normal, ✓ AAA · large.
#2563eb on #ffffff: 5.17:1 → ✓ AA · normal, ✓ AA · large, ✕ AAA · normal, ✓ AAA · large.
#ffffff on #d6f64b: 1.22:1 → ✕ AA · normal, ✕ AA · large, ✕ AAA · normal, ✕ AAA · large.
My separate calculation matched the one the app shows, to two decimals, in 5 of 5 pairs.
I didn't measure times: it's an arithmetic calculation that updates when the color changes, and the latency of my measuring tool is greater than that of the calculation.
What went wrong or what limit I saw: The checker compares two flat colors: it doesn't measure text over photos or over gradients. In this test I didn't use the color-blindness simulation or measure times. The edge case is useful: #777777 on white (4.48:1) fails AA for normal text and #767676 (4.54:1) passes, by a very small margin.
Tested by Gabriel Maglia on with PhotoCut (web) in Chrome 154.0.8037.98 · macOS 14.8.8 · PhotoCut version: commit bc142c7.