'Slightly Warmer' Is Not a Color Agreement — How to Specify Color for Your Supplier with L*a*b* and ΔE

Ask three people for "a warm beige" and you get three different colors. Ask your supplier for it and you get a fourth. As long as color is agreed in words, every agreement is open to interpretation — and the difference only comes to light once the delivery is already in. The solution is not to describe color but to specify it: in numbers that supplier, purchasing and the inspector on the floor all read identically.

The Three Numbers: L*, a* and b*

The international system for this is CIELAB, standardized in CIE 15:2018 and ISO 11664-4. Every color gets three coordinates. L* is the lightness, from 0 (black) to 100 (white). a* runs from green (negative) to red (positive). b* runs from blue (negative) to yellow (positive). Together, these three numbers fix every conceivable color at one specific location in a three-dimensional space.

In practice you often work with the derived CIELCh notation, which lies closer to the way we name color: L* for lightness, C* for chroma (saturation) and h for the hue angle. It is the same space, only expressed in polar coordinates instead of rectangular ones. Where "warm beige" leaves room for discussion, L* 74, C* 12, h 78 does not — that is one point, without margin for interpretation.

Where the Real Work Lies: ΔE

A color coordinate fixes the target color. But a supplier never delivers exactly that color — there is always a deviation. So the question is not "which color", but "how much deviation is permitted". That is what ΔE measures: the distance between your standard and the sample in the color space, expressed in one number.

That single number is useful for a quick pass/fail, but it hides where the deviation lies. That is why ΔE breaks down into components you can control separately: ΔL* (the lightness difference), ΔC* (the saturation difference) and ΔH* (the hue difference). This way you see not only that a sample deviates, but also in what respect — and that is often decisive. A deviation in lightness stands out differently from a deviation in hue, and for some products a shift in chroma is acceptable while a shift in hue is not. Limits on the individual components give you that control; a single total ΔE does not.

The formula with which you calculate ΔE also matters. The older ΔE*ab measures the bare Euclidean distance in the CIELAB space, but that space is not perceptually uniform: an equally large arithmetic distance looks like a larger difference in one region than in another. The modern formula CIEDE2000 (ISO 11664-6) corrects for this with weightings for lightness, chroma and hue, and therefore correlates better with what the eye actually sees — especially with small differences and with saturated colors. So agree not only on a ΔE value, but also on which formula you apply; the same sample yields different numbers under ΔE*ab and CIEDE2000.

Why measure and not just look? Because the eye becomes unreliable at small differences. At a ΔE below 0.5, even top experts hardly perform better than guessing — 50 to 60 percent correct decisions. The eye remains indispensable for context and borderline cases, but for hard acceptance, numbers are more reliable.

The Pitfall: A ΔE Only Applies Under Agreed Conditions

This is where things go wrong in practice. A ΔE number is not absolute — it only applies under the measurement conditions under which it was determined. Three things determine whether your numbers are comparable at all.

First, the measurement geometry. If you measure with a 45°/0° geometry and your supplier with a d/8 sphere geometry, you are in fact measuring something different — certainly with glossy or textured surfaces, where including or excluding the gloss (SCI/SCE) shifts the values. Record the geometry explicitly.

Second, the light source. If you measure under D65 and your supplier under another source, you are comparing apples with oranges, because the color values are partly determined by the chosen light source. Agree on the light source (and the standard observer, 2° or 10°).

Third, metamerism: two materials can measure identically under D65 and visibly differ under store fluorescent or incandescent light. A single ΔE under one light source can thus reassure you while the customer sees the difference plainly in the store. That is why you test critical colors under multiple light sources — typically D65, TL84/F11 and incandescent A.

What You Record with Your Supplier

A watertight color agreement contains five things: the standard (which master applies), the color space and formula (CIELAB with CIEDE2000), the tolerance — not only a total ΔE but also limits on ΔL*, ΔC* and ΔH* — the measurement conditions (geometry, light source, observer), and an agreement on metamerism for colors where that risk plays a role.

Choose the tolerance deliberately: too tight leads to unnecessary rejection and costs, too loose to complaints. The right value depends on your product, on how critical the eye is for that color, and on where the color will sit next to another — a difference that is invisible on its own jumps out as soon as two parts sit against each other.

What This Gives You

Record your color in L*a*b*, with a ΔE tolerance including component limits, measured under agreed conditions, and you have a specification that is not open to interpretation. No more "slightly warmer", but a number that holds up in a specification document, a purchase order and an incoming inspection.

If you want to master this — visual assessment and instrumental measurement, and aligning the two — the Color Assessment program is built for it. For quick measurements in the field, the ColorReader Pro offers an accessible instrument, and for setting up your color process you can engage our consultancy.

---

Source: Kotterink, M. (2025). Kleurbeoordelen in de Praktijk — Deel 2: Kleurmeting en beoordeling. Uitgeverij SNKI. With the standards applied therein (CIE 15:2018, ISO 11664-4/-6, ISO 3664) and the work of McLaren (1976) on the limits of visual assessment.