Color measurement guide
How Delta E measures perceptual color difference
Delta E turns a pair of color samples into a numerical difference. This guide explains what CIEDE2000 actually compares, how to read the result, and where the number stops being enough.
Delta E color difference
What does Delta E measure?
Delta E measures the numerical difference between two colors in a color space designed to relate more closely to visual perception than raw RGB values. This article focuses on CIEDE2000, written ΔE00, which starts with CIELAB coordinates and applies weighting and interaction corrections for lightness, chroma, and hue. A lower result means the pair is closer under that formula. It does not guarantee that every observer, display, material, background, or workflow will treat the difference identically.
The short version
- Name the formula: Delta E 76, CIE94, and CIEDE2000 do not produce interchangeable numbers.
- Read the value with the visible samples and the viewing or production conditions.
- Use HSB channel differences to diagnose an edit, not as a replacement for CIEDE2000.
- Use a separate relative-luminance contrast method for WCAG decisions.
From HEX values to a perceptual comparison
A HEX color is a compact notation for red, green, and blue display channels. It is excellent for sharing a web color, but the coordinate spacing is not designed so that every equal numerical step looks equally large. A change of ten code values in a dark blue may not feel like the same change in a bright yellow. Before a perceptual difference formula can be used, the samples must be parsed consistently and transformed through a defined color-management path into coordinates such as CIELAB.
CIELAB represents lightness with L* and opponent color dimensions with a* and b*. It is more suitable for difference work than raw RGB, but a simple Euclidean distance in Lab, commonly associated with Delta E 76, still does not model every region equally well. Later formulas added adjustments based on visual data. CIEDE2000 is one of those refinements. The official ISO/CIE 11664-6:2022 publication describes it as an extension of the CIE 1976 L*a*b* difference with corrections intended to improve correspondence with perceived relative differences.
What CIEDE2000 changes
CIEDE2000 does not merely subtract three Lab coordinates. It calculates lightness, chroma, and hue differences, applies weighting functions to those components, and includes an interaction term between chroma and hue. The formula also handles hue as an angle, which requires care around the zero-degree boundary. These steps respond to known nonuniformities in color spaces and to the fact that the same coordinate movement can be experienced differently depending on the region and direction of the color change.
The implementation details matter. Gaurav Sharma, Wencheng Wu, and Edul Dalal published implementation notes and supplementary test pairs because apparently reasonable code can disagree at angular boundaries or special cases. Their maintained resource includes reference data for checking software. Toon Tone uses the established CIEDE2000 implementation in its color library rather than retyping the formula inside the page. The calculator then adds visible HEX and HSB context so the output is easier to inspect, while keeping those diagnostics separate from ΔE00 itself.
Why formula names and parameters matter
The phrase ‘Delta E’ names a family of difference calculations rather than one universal equation. Delta E 76 uses a direct Lab distance. CIE94 and CIEDE2000 introduce different corrections, and other formulas exist for particular applications. If two tools use different versions, their results can disagree without either interface necessarily being broken. A screenshot that shows only ‘Delta E: 2.4’ omits a critical part of the measurement description. A reproducible report should say which formula, inputs, color assumptions, and parameters were used.
CIEDE2000 also includes parametric factors often represented as kL, kC, and kH. Reference conditions commonly use values of one, as this browser calculator does through its library defaults, but specialized workflows may choose parameters defined by their standards. A number from those workflows should not be compared casually with a default browser result. If brand, textile, paint, laboratory, or manufacturing rules specify a formula and tolerance, follow the full specification. The web tool is best for learning, digital triage, and reproducible comparisons under its stated assumptions.
How to interpret a Delta E result without overclaiming
The safest universal reading is directional: zero is an identical calculated pair, and larger values represent larger differences under the selected formula. Practical ranges can help a person sort results, but they do not create a single just-noticeable threshold for every situation. Observers vary. Samples shown adjacent are easier to compare than samples separated by time or space. Size, texture, gloss, illumination, adaptation, surrounding colors, display profile, and calibration can all influence the experience that the formula is being asked to approximate.
Use a threshold only when it belongs to the relevant workflow. A design team might define an internal tolerance for checking token migrations on calibrated reference displays. A printer may work from measured physical samples under specified lighting. A game can map Delta E into a score chosen for feedback and entertainment. Those three uses can be valid while having different acceptance rules. When no governing rule exists, describe the result cautiously, show both swatches, record the inputs, and decide whether the observed change affects the actual task.
How HSB differences add an editing explanation
HSB organizes a display color into hue, saturation, and brightness. These controls match how many people edit a remembered shade: choose the family, tune intensity, then adjust light or dark. After CIEDE2000 says that two samples differ, the HSB deltas can indicate the most obvious direction of change. A large brightness gap suggests a value correction; a large saturation gap suggests that one sample is more muted; a hue gap shows movement around the color wheel. Circular hue distance avoids reporting a huge jump from 359 to 1 degrees.
HSB itself is not a perceptually uniform measurement space. A ten-degree hue move can look different depending on saturation and brightness, and a ten-point brightness change does not have a constant perceptual effect across colors. That is why the calculator does not combine the three HSB differences into a second competing score. It presents them as diagnostics next to CIEDE2000. This pairing is useful in Toon Tone because a player can understand both overall closeness and which familiar slider is most worth correcting in the next round.
A five-step Delta E workflow
Use this checklist when you need a comparison that another person can understand and repeat.
1. Capture exact inputs
Use the original HEX or measured coordinates and record any relevant color profile instead of relying on a compressed screenshot.
2. Name the formula
State CIEDE2000 or the actual alternative. Do not report an unlabeled ‘Delta E’ as though every version were interchangeable.
3. Compare in context
View the real size, background, medium, theme, and lighting conditions that matter to the decision.
4. Diagnose the direction
Use HSB deltas or source-channel values to identify the edit while keeping them separate from the perceptual score.
5. Apply the governing rule
Use the project, accessibility, laboratory, print, or game rule that actually owns the pass/fail decision.
CIE · University of Rochester · W3C
Primary references and implementation material
These sources define the formula context, provide implementation checks, and document the separate WCAG contrast method used for web accessibility decisions.
Put the explanation into practice
Delta E color difference FAQ
Is CIEDE2000 the same as every Delta E value?
No. Delta E 76, CIE94, CIEDE2000, and other formulas use different calculations. Always record the version with the number.
What does a Delta E of zero mean?
It means the two processed samples are identical under the selected calculation. It does not prove that the source devices or physical objects were measured correctly.
Can one Delta E threshold work for all projects?
No. Acceptance depends on formula, viewing conditions, materials, observers, risk, and the standard governing the task.
Why show HSB if CIEDE2000 uses Lab?
HSB gives a familiar editing diagnosis. It helps identify hue, saturation, or brightness drift but is not used as a replacement for ΔE00.
Can Delta E prove WCAG contrast?
No. WCAG contrast uses relative luminance and defined success criteria. Run the appropriate contrast evaluation separately.