What it does
Conservators, colour scientists and print and textile labs measure colour with a spectrophotometer or colorimeter and compare readings over time: has a dyed textile faded on display, has a varnish yellowed, did a cleaning test change the paint? A single number for “how different” is the colour difference, ΔE. This calculator gives the three CIE versions side by side:
- ΔE*ab (1976): the straight distance between the two points in CIELAB. Simple, but it overstates differences in strong colours.
- ΔE94: weights chroma and hue differences by the chroma of the reference colour, so it is not symmetric: the before reading is the reference.
- ΔE00 (CIEDE2000): the current CIE recommendation, with extra corrections for blue hues and near-greys. The parametric factors kL, kC and kH let you weight lightness, chroma and hue; 1, 1, 1 is the reference setting.
It also splits the change into ΔL* (lighter or darker), ΔC* (more or less saturated) and ΔH* (hue), and says in words what happened, which is often more useful than the total.
How to use it
- Choose what your numbers are: CIELAB, LCh, XYZ (with Y = 100 for the white) or sRGB (0 to 255, or hex codes). Switching converts the two colours on screen.
- Pick the reference white. It only matters when converting from XYZ or sRGB; for Lab* or LCh values ΔE does not depend on it.
- Type the before (reference) and after colours. The swatches give a rough screen preview.
- For a survey, paste one object or measuring spot per line into Fading survey: a name, then three before and three after numbers, or press Open a file for a CSV or Excel file. Set the flag level your institution uses.
Worked example
The page opens with a made-up red silk band measured before and after a display: L*, a*, b* = 42.1, 48.3, 22.6 before and 44.0, 44.9, 21.8 after. It shows ΔE00 = 2.07, ΔE94 2.19 and ΔEab 3.98. ΔL is +1.90 and ΔC* is −3.41: the colour got lighter and less saturated, which is what fading looks like. ΔE*ab is almost twice ΔE00 because the red is strongly saturated, where the 1976 formula overstates changes.
The example survey has six made-up samples. With the flag level at 2, the summary reads “6 samples: mean ΔE00 1.80, largest 3.53 (Green wool leaf); 3 at or over 2”. The yellow silk thread has the biggest ΔE*ab (7.58) but only ΔE00 2.20, again because most of its change is chroma in a very saturated colour. The black outline barely moved (ΔE00 0.39).
Tips
- Measure the same spots each time (a template or photographs help), with the same instrument, aperture, illuminant and observer, and specular component included or excluded the same way.
- Average several readings per spot. Differences under about 1 ΔE00 are often within the repeatability of hand-held instruments.
- Use the CSV download to keep the record; it has all three formulas and the components.
FAQ
Which ΔE should I report? CIEDE2000 for new work, and say which kL, kC and kH you used. Give ΔE*ab as well if you compare with older records that used it.
Why are the swatches not exactly right? Screens are not calibrated, and some measured colours lie outside what an sRGB screen can show; the page says when a swatch is clipped.
Is my data uploaded? No. Everything runs in your browser and keeps working offline.
Where the numbers come from
- CIELAB, ΔE*ab and ΔE94: CIE 15 (Colorimetry) and CIE 116 formulas. CIEDE2000: G. Sharma, W. Wu and E. N. Dalal (2005), Color Research and Application 30, 21-30, whose published test pairs are used in fibub's tests. sRGB: IEC 61966-2-1 matrix and transfer curve; Bradford chromatic adaptation.
- Checked against colour-science 0.4.7 (Python) on 2,000 random colour pairs: ΔE76, ΔE94 (both weightings), CIEDE2000 with random kL, kC and kH, LCh, and sRGB to Lab under D65 and D50, all within 10⁻¹².
- The fading survey example is made up for this tool; it is not a measurement of any real object.
Standard scientific values, such as constants, atomic weights and melting points, are facts found in every textbook; we name where ours were checked. If you are an author or publisher and want a value, a credit or a table changed or removed, tell us and we will act quickly.
Last updated 2026-11-09.