Field note · Flexo prepress

What is Delta E? Color tolerances for flexo printing

What Delta E measures, why ΔE76 and ΔE2000 give different numbers, published G7 and ISO tolerances, how M0 and M1 measurement change results, and which color data belongs on the flexo job record.

Plate area · jobs · customers · AR — one system of record
Flexoworks
  • flexo
  • delta e
  • color difference
  • color tolerance
  • CIEDE2000
  • spot color
  • color management

Delta E (ΔE) is the distance between two colors in CIELAB space: the target and what was printed. The bigger the number, the bigger the visible difference. On its own, a Delta E number tells you little. "ΔE 2" only means something once you also name the formula, the measurement condition, and the reference it was measured against. Most color disputes between brand, trade shop, and printer start because one of those three was never written down.

What Delta E measures

A spectrophotometer reports a color as L* (lightness), a* (green to red), and b* (blue to yellow). The original 1976 formula, ΔE76 or ΔE*ab, is plain geometry: the square root of ΔL*² + Δa*² + Δb*². Later formulas weight those differences to better match how people see them.

There is no universal pass mark. ChromaChecker's review of the research found that a ΔE of 2 to 3 is almost universally accepted for spot and brand colors, 8 or more is almost universally rejected, and everything between 3 and 8 depends on who is looking. Tighter tolerances get harder, and more expensive, to hold.

ΔE76, ΔE94, and ΔE2000 give different numbers

CIE 1976 is standardized as ISO/CIE 11664-4 and CIEDE2000 as ISO/CIE 11664-6. CIEDE2000 (ΔE00) forgives chroma differences in saturated colors, where the eye is less sensitive, and is stricter on near-neutrals, where it is more sensitive. The same pair of measurements can pass one formula and fail another. These three pairs come from the CIEDE2000 test data published by Sharma, Wu, and Dalal:

Color pair ΔE76 ΔE94 ΔE2000
Saturated blue (pair 1) 4.00 1.40 2.04
Deep red (pair 30) 3.89 1.42 1.41
Near-neutral gray (pair 7) 2.24 2.24 2.37

ΔE76 and ΔE94 are our calculations from the published L*a*b* values. A "ΔE 3" spec rejects the red under ΔE76 and passes it easily under ΔE2000. The gray goes the other way. G7 now uses ΔE2000, and ISO 12647-7:2016 moved proof tolerances from the 1976 formula to ΔE2000 with new limits, because the values cannot be converted directly.

Published tolerances

Specification What is checked Tolerance
G7 Targeted, press Substrate / CMY solids / K solid / RGB overprints ΔE00 ≤ 3.0 / 3.5 / 5.0 / 4.2
G7 Colorspace, press All IT8.7/4 patches Average ΔE00 ≤ 3.5, 95th percentile ≤ 5.0
G7 Colorspace, proof All IT8.7/4 patches Average ΔE00 ≤ 1.5, 95th percentile ≤ 3.0
G7 Grayscale Weighted ΔL* and ΔCh on gray scales Average ≤ 1.5, maximum ≤ 3.0
ISO 12647-7:2016 proof Spot color solids ΔE00 ≤ 2.5

The G7 figures come from Idealliance's G7 Master pass/fail requirements (v35). Idealliance calls the grayscale limits the bare minimum and tells serious printers to aim for about half. Its flexo annex makes one allowance for legacy platemaking: gray balance errors under 20% dot are ignored, solids are judged the same as any other process, and three printed samples may be averaged. Flexo shops using modern screening, such as the hybrid screens in our AM and FM note, do not need the annex.

For brand spot colors in packaging, the brand owner's spec usually governs, and it is often tighter for key colors. ISO 12647-6:2020, the flexo part of the series, also expects files for flexo to come with an ISO 12647-7 proof unless the parties agree otherwise.

The measurement condition changes the number

ISO 13655 defines four instrument modes. M0 is legacy, with an unspecified amount of UV. M1 approximates D50 daylight, UV included. M2 cuts UV out. M3 adds a polarizing filter. Many white boards and label stocks carry optical brighteners, which turn UV into visible blue. Two M0 instruments can read the same sample differently, so an inline system can pass a job that the customer's handheld then rejects at press side. The FTA recommends M1, and G7 auditors measure in M1 unless agreed otherwise. ChromaChecker warns that picking a Pantone library in the wrong mode (M0, M1, or M2) can easily mean an accidental 5 ΔE difference before a drop of ink is printed.

The reference has to be fixed too

A Pantone number is not a target. Swatch books vary by edition and age, and a printed sample fades. A numeric L*a*b* target is better, but it is valid for one illuminant only. The strongest reference is spectral data in CxF/X-4 (ISO 17972-4), which can carry the solid, its tints, and its overprints. It is still only valid for the ink and substrate it was measured on.

When a spot color misses, check the cheap causes before the plate: substrate color and surface treatment, anilox volume and ink film, and ink formulation. Tints that miss while the solid is on target usually point to the dot gain curve. If the job is moving spot colors into a fixed palette, the extended color gamut note covers how close a 7-color build can get.

What to keep on the job record

Record the reference (Pantone name plus library edition, or the CxF file), the target L*a*b*, the formula, the tolerance, the measurement condition, the instrument model and backing, the substrate and ink set, the measured value on the approved proof and press OK, and who signed off. For a trade shop, that record settles reruns. When a color is challenged six months later, it shows whether the target, the substrate, or the measurement changed, before anyone pays for a plate remake.

Common questions

What is Delta E in printing?

Delta E (ΔE) is a number for the color difference between a target and a printed sample, calculated from their CIELAB (L*a*b*) values measured with a spectrophotometer. The larger the number, the more visible the difference. It only means something when the formula (usually ΔE2000), the measurement condition (usually M1), and the reference are also stated.

What is an acceptable Delta E for printing?

There is no single number. Research summarized by ChromaChecker found ΔE 2 to 3 almost universally accepted for brand and spot colors and 8 or more almost universally rejected. G7 Targeted allows ΔE2000 of 3.5 on CMY solids and 5.0 on black, and ISO 12647-7:2016 allows 2.5 on spot color solids in a contract proof. For packaging, the brand owner's specification usually sets the limit.

What is the difference between Delta E 76 and Delta E 2000?

ΔE76 is the straight-line distance between two colors in L*a*b* space. ΔE2000 weights lightness, chroma, and hue differences to match human vision more closely, so it is more lenient on saturated colors and stricter on near-neutrals. The same pair of colors can give very different values: one published blue pair measures 4.00 in ΔE76 and 2.04 in ΔE2000.

Should color be measured in M0 or M1?

M1 is the recommended mode for most graphic-arts work. ISO 13655 defines M1 as D50 with a specified UV content, so instruments from different makers agree on substrates with optical brighteners. M0 is a legacy mode with an unspecified UV content, and two M0 instruments can read the same sample differently. G7 auditors measure in M1 unless agreed otherwise.

Built for flexo trade shops

Keep production context with the job.

See how Flexoworks connects customers, prepress, plate production, delivery, and plate-area billing.