Short answer
ΔE is a single number for the difference between two colours, calculated from their CIELAB values. ΔE76 is the straight-line distance in CIELAB; ΔE94 and CMC shrink chroma and hue differences for saturated colours; CIEDE2000 adds corrections for lightness, hue and the blue region so that equal numbers look more nearly equal. The same pair of colours can therefore score 4.0 in ΔE76 and 2.0 in CIEDE2000; a ΔE is meaningless without the formula.
CIELAB was designed in 1976 to be roughly perceptually uniform, so the obvious difference measure is Euclidean distance: the square root of the summed squares of ΔL*, Δa* and Δb*. It is simple, symmetric and still used in many standards and tools. Its weakness is that CIELAB is not uniform: people are much less sensitive to chroma differences in saturated colours than in near-neutrals, sensitivity to hue varies around the hue circle, and blues behave oddly. So ΔE76 overstates differences between saturated colours (a yellow that looks fine scores 5) and understates some near-neutral ones (a grey tinted slightly pink scores 2 but is obvious).
The fix is to express the difference as lightness, chroma and hue components and divide each by a weight that grows where people are less sensitive. The CMC(l:c) formula, from the Society of Dyers and Colourists' Colour Measurement Committee in 1984, does this with weights depending on lightness, chroma and hue angle, plus parameters l and c: 2:1 is the usual acceptability setting, allowing twice as much lightness difference, and 1:1 the perceptibility setting. CIE94 (CIE 116-1995) is a simpler CIE version with chroma-dependent weights and parametric factors for different industries — graphic arts and textiles use different constants. Both are asymmetric: the reference's chroma sets the weights, so swapping sample and reference changes the result slightly.
CIEDE2000 (CIE 142:2001) keeps the CIE94 idea and adds corrections the older formulas missed: a lightness weight that is smallest around mid-grey and grows towards light and dark, a hue weight that varies with hue angle, a rescaling of a* for near-neutral colours, and a rotation term that couples chroma and hue differences in the blue region, where the perceived tolerance ellipses are tilted. It was fitted to several combined sets of visual judgements. It is harder to implement — Sharma, Wu and Dalal published test data precisely because many early implementations were wrong — but it is now the formula in ISO 12647-7 and most modern tools, including Colourwise's.
The table scores seven pairs with each formula using Colourwise's implementations; the first is the published CIEDE2000 test pair 1, which reproduces its published value of 2.0425. The pattern is instructive. For the saturated yellow and blue pairs ΔE76 is two to five times the weighted results. For the reddish tint on mid-grey, CMC and CIEDE2000 score higher than ΔE76, because near-neutral chroma differences are visible. For equal lightness steps, CIEDE2000 scores the mid-grey step highest and the dark and pale steps lower, while ΔE76 and ΔE94 treat all three the same. None of the four is 'the real difference'; they are models of average judgements under reference conditions.
| Pair (CIELAB) | ΔE76 | ΔE94 (graphic arts) | CMC 2:1 | ΔE00 |
|---|---|---|---|---|
| Saturated blue, small hue/chroma shift (Sharma et al. pair 1) | 4 | 1.4 | 1.74 | 2.04 |
| Strong yellow, 5 units less b* | 5 | 0.97 | 1.51 | 0.99 |
| Red, hue turned towards orange | 6.4 | 2.85 | 4.07 | 3.35 |
| Mid grey, 2 units lighter | 2 | 2 | 0.92 | 1.99 |
| Mid grey, 2 units towards red (a*) | 2 | 2 | 3.13 | 2.81 |
| Dark grey, 2 units lighter | 2 | 2 | 1.65 | 1.4 |
| Pale grey, 2 units lighter | 2 | 2 | 0.7 | 1.24 |
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
StandardStrong evidence
CIE94 (CIE 116-1995) extends the CIE 1976 colour-difference formula with a correction for chroma-dependent variation in perceived colour difference, reference viewing conditions and parametric factors for conditions of use.
Source: CIE 116-1995 Industrial colour-difference evaluation
StandardStrong evidence
CIEDE2000 (CIE 142:2001) retains the CIE94 corrections and adds further corrections for lightness, chroma and hue dependence derived from experimental data, including a term for the blue region.
Source: CIE 142:2001 Improvement to industrial colour-difference evaluation; The development of the CIE 2000 colour-difference formula: CIEDE2000
FactModerate evidence
The CMC(l:c) formula was published in 1984 for the Colour Measurement Committee of the Society of Dyers and Colourists; a 2:1 ratio is commonly used for acceptability and 1:1 for perceptibility.
Source: Modification to the JPC79 colour-difference formula; Color difference (Wikipedia)
Colourwise analysisStrong evidence
Sharma, Wu and Dalal's CIEDE2000 test pair 1 (L* 50, a* 2.6772, b* −79.7751 against L* 50, a* 0, b* −82.7485) has a published ΔE00 of 2.0425, which Colourwise's implementation reproduces, while its ΔE76 is 4.00.
Based on: Computed with the site's colour-difference functions and checked against the published test value.
Reviewed 29 September 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.