A neutral reading interface

Use black or charcoal for reading text and white for the canvas. Mid gray and silver need individual checks for captions and field boundaries.

Colors and palette cases

Editorial example reviewed September 13, 2026. Numeric results below are calculated from these values.

Nearest reference name

Black

RGB 0, 0, 0

Explore color

Nearest reference name

White

RGB 255, 255, 255

Explore color

Nearest reference name

Gray

RGB 128, 128, 128

Explore color

Nearest reference name

Silver

RGB 192, 192, 192

Explore color

Nearest reference name

Charcoal

RGB 64, 64, 64

Explore color

Assign roles and check the pair

The lightest and darkest supplied values are the starting reading pair. The accent label adds black or white, whichever has greater contrast. You can change the accent below.

Accent surface

A palette in context

Use this preview to inspect reading text and an accent label.

Example action

Reading text #000000 on #FFFFFF: 21.000:1 - Pass normal text AA.

Accent label #FFFFFF on #000000: 21.000:1.

Checks use unrounded ratios and opaque sRGB. The example action is a visual sample. Test boundaries, focus and hover in your actual component.

All color pairs: normal text AA

Rows are text; columns are backgrounds. Each cell gives the ratio and AA result.
Text / background	#000000	#FFFFFF	#808080	#C0C0C0	#404040
#000000	1.00:1 Fail	21.00:1 Pass	5.32:1 Pass	11.54:1 Pass	2.03:1 Fail
#FFFFFF	21.00:1 Pass	1.00:1 Fail	3.95:1 Fail	1.82:1 Fail	10.37:1 Pass
#808080	5.32:1 Pass	3.95:1 Fail	1.00:1 Fail	2.17:1 Fail	2.63:1 Fail
#C0C0C0	11.54:1 Pass	1.82:1 Fail	2.17:1 Fail	1.00:1 Fail	5.70:1 Pass
#404040	2.03:1 Fail	10.37:1 Pass	2.63:1 Fail	5.70:1 Pass	1.00:1 Fail

:root {
  --surface: #FFFFFF;
  --text: #000000;
  --accent: #000000;
  --on-accent: #FFFFFF;
}

Edit and export in Studio

Keep the scope of each result clear

All comparisons use opaque sRGB and WCAG relative luminance. Normal-text AA requires 4.5:1; a displayed rounded ratio never overrides the unrounded pass or fail. A palette is not a complete accessibility audit.

W3C contrast requirements ? Methods and limitations ? Turn values into semantic tokens