Skip to main content
BVDART Logo
BVDART
← All guidesColor

Just-noticeable difference in color

A just-noticeable difference is a change an observer can detect under specified viewing conditions. It is an experimental threshold, not a universal property of a hex color.

Approach a visible difference, then reveal its location.

Surround
Displayed ΔE76 1.173#808080 → #838383 · Move slowly, observe, then reveal.

One patch changes lightness; the other matches its background. Record your own observation as size and surround change. This is an exploration on your display, not a calibrated threshold test; rounding can make different requested steps identical.

A small lightness difference slowly emerges. Its location is revealed after a pause for you to look. Use any control to take over.

What to try
  1. Choose Start at zero and increase ΔL* slowly without revealing the target.
  2. Record I notice a difference, then compare another Patch size or Surround.
Open in the Lab ↗

What to look for: Your observation depends on the viewing conditions. The displayed hex pair can stay identical across several small requested steps.

How it works

Patch size, surrounding colors, adaptation, display calibration and the observer all affect detectability. A color-difference formula predicts aspects of perception, but it does not measure whether a particular visitor can see a change on their current screen.

Here, a small change in Lab lightness produces a second displayed patch. The browser must convert and round that color back to eight-bit sRGB. At very small steps, several requested values can therefore produce the same visible triplet.

Use it in the Lab

When judging a palette or a subtle adjustment, compare at the intended viewing size. Keep numerical readouts as evidence alongside the image, especially when reducing colors.

Open in the Lab ↗

Limits & distinctions

This is an exploratory comparison, not a calibrated visual test. Avoid treating a fixed ΔE threshold as a guarantee of invisibility.

Further reading

The Just Noticeable Difference is a psychophysical concept defining the smallest change in a stimulus (here, color) that a human observer can detect. In color science, it provides the practical threshold below which two colors are effectively indistinguishable.

Delta E (ΔE) and the JND

The CIELAB Delta E (ΔE) metric was designed so that a value of 1.0 roughly corresponds to one JND under controlled laboratory conditions. In practice, the thresholds are commonly interpreted as:

  • ΔE < 1.0 — Not perceptible to most people
  • ΔE 1–2 — Perceptible through close observation
  • ΔE 2–10 — Perceptible at a glance
  • ΔE > 10 — Obviously different colors

CIE76 Formula

In CIELAB: ΔE⁶ = √((L₂–L₁)² + (a₂–a₁)² + (b₂–b₁)²) This is simply Euclidean distance in L*a*b* space. More advanced formulas (CIE94, CIEDE2000) apply correction terms to better match human perception in problematic hue regions.

Practical Application: Color Grouping Thresholds

Our palette extraction uses a grouping threshold of 32 in RGB Euclidean distance. This is deliberately much larger than the JND because the goal is to merge visually similar colors, not identify barely-distinguishable ones. A JND-based threshold would produce a palette with too many entries to be useful.

However, JND is critical for quality control: if a printed reproduction has a ΔE of less than 1.0 compared to the original, the colors are considered a perfect match.

Key Takeaway

The JND establishes the absolute floor of meaningful color difference. Any color operation — quantization, gamut mapping, compression — that stays within ΔE < 1.0 is, by definition, lossless from a perceptual standpoint.

Related Lab categories

Color Theory Lab ↗

Sources & references

Guide updated