Smaller colour patches need larger hue differences to stay discriminable

What it is

Size-dependent color discriminability means that a color pair separable in large swatches may not remain separable when rendered as scatterplot points, thin lines, or narrow bars. As a mark subtends a smaller visual angle, a larger color difference is generally needed to maintain the same detection probability. Although the title says “patch area,” diameter, stroke width, length, shape, spacing, and viewing distance also matter; geometries with the same pixel area are not interchangeable.

Why it happens

Chromatic and spatial vision jointly constrain small marks. Small marks provide less chromatic signal, while antialiasing, display subpixels, edge mixing, and the surrounding background occupy a larger share of the rendered stimulus. Dense layouts add crowding and search interference. Size also affects differences along hue, lightness, and chroma axes differently. CIELAB and related differences defined for standard viewing conditions do not automatically predict arbitrary tiny marks. “Equal color difference” therefore describes distance in a specified model, not guaranteed discriminability in a finished chart.

Studying it

Mark-perception experiments vary size and color difference across points, lines, bars, and regions, asking participants to detect a difference or find a category and fitting detection probability rather than reporting one mean accuracy. To test the title's hue claim, use hue pairs matched in lightness and chroma and measure category confusion; evidence about mark-dependent overall color difference does not by itself establish a fixed hue-angle increment. Record color coordinates, background, geometry, edge rendering, pixel density, viewing distance, and display environment. Crowdsourcing samples realistic devices but sacrifices calibration and visual-angle control. Validate a fitted model on held-out colors, sizes, and tasks; do not transfer a threshold from one mark geometry directly to another.

Where it stops holding

The title gives a general direction, not a fixed minimum size, color difference, or category limit. Very high contrast, few categories, direct labels, and slower lookup can mitigate small-size losses; low vision, color-vision variation, mobile viewing, and compressed screenshots can amplify them. Large filled regions are not automatically safe: thin shapes, holes, intricate boundaries, and neighboring backgrounds change effective size. Lightness contrast can provide redundancy, but hue, lightness, and chroma should not be collapsed into “more vivid.”

Applying it

• Test palettes inside the final chart at their real diameters, stroke widths, lengths, and backgrounds. Large legend swatches cannot certify small marks.
• For small marks, reduce simultaneous categories and increase measured hue and lightness separation. Add shape, outlines, labels, or filtering for consequential classes.
• Record target breakpoints, pixel density, and export path. Inspect antialiasing and color after zoom, screenshots, projection, and print.
• Estimate confusion matrices and intervals with category search and matching tasks. Replace a color or encoding when a pair is unstable on critical marks; do not impose one universal error cutoff.

Related

• Same group: U1.05.1 Hue is a categorical channel with no inherent order · U1.05.2 Encoding a continuous variable as hue is read as discrete categories · U1.05.4 Adjacent colours bias each other's hue · U1.05.5 Culturally loaded hues override the assigned mapping
• Adjacent: U1.09.4 Discriminable-level limits drop as marks shrink · U1.06.5 Saturation differences are nearly invisible on small marks
• Search terms: mark-aware color difference · color discriminability · visual angle