Laden...
Laden...
The spectrophotometer approves the batch. The number falls within tolerance, the report is green. And yet the operator holds the sample up to the light, frowns, and says: something is off. Who is right?
You would expect the instrument to lead — after all, it measures objectively, and the eye is only human. But precisely in this case, the eye is right surprisingly often. And that is neither coincidence nor stubbornness. It says something about what a meter does and does not do.
A meter is not a lie detector for quality. It answers a very precise question — and everything outside that question it does not see.
To begin with, an instrument measures under one defined light source and for one standard observer. Two surfaces that measure identically there can visibly fall apart as soon as the color is used under a different light source. That is metamerism, and the eye that assesses the color in the real application catches exactly what the measurement left behind at the measurement condition. The meter was right — for its own light source. The eye is right for the place where the product will eventually hang.
Add to this that a meter measures a spot, while the eye sees the whole. A spectrophotometer reads a small area; a deviation concentrated on an edge, a seam or one side of a panel disappears into the average. The eye sees streaking, clouding, a gloss difference or a texture deviation that hardly counts numerically, but stands out immediately in the real product.
And then there is the difference that color researcher Paul Green-Armytage articulates sharply: a meter records a spectral profile or a psychophysical value — a number. The eye assesses the appearance in context: gloss, direction, the surrounding colors, the way the light falls. Two things that measure the same do not have to look the same. Moreover, a deviation can fall neatly within your Delta E tolerance and still be visible, simply because the eye is more sensitive in certain color regions than a single numerical threshold suggests.
Not always — and that is where the skill lies. If one assessor on a bad day sees "something", that is no reason to throw out a green measurement report. But if several trained assessors consistently see the same thing the meter misses, that is not noise. It is a signal that your measurement is answering the wrong question.
The usable decision rule is therefore not "the meter wins" or "the eye wins", but: let the instrument establish whether something deviates measurably, and let the trained eye establish whether it matters in the application. If the two contradict each other, do not treat that as an error by the assessor, but as a clue — almost always pointing to the measurement light source, the measurement geometry, the measured spot, or the reference you are working against. The conflict is the diagnosis, not the problem.
There is a hard condition underneath this whole story. The eye may only be given the final say if it is demonstrably reliable. "I see something" is not an argument from someone whose discrimination ability you do not know — and as we described earlier, color perception varies strongly from person to person and some people are not even aware of their own limitation.
That is why it only works if you know who is assessing. Map the color discrimination ability of your assessors objectively — with an assessment such as ColorAptitude — so that a visual judgment contradicting the meter carries weight because it comes from the right eyes. A trained eye that says "fail" is an instrument. A random eye that says "fail" is an opinion.
The conflict between meter and eye becomes useful the moment you read it as a diagnosis instead of a difference of opinion. Three causes cover virtually all cases, and each points to a different intervention.
If the cause is metamerism, the sample matches under the measurement light source but not under the light source of the application. The intervention is then not measuring more strictly under D65, but measuring under multiple light sources — typically daylight, a store fluorescent and incandescent — and comparing the reflectance curves. If the cause lies in the spot measurement, the deviation is concentrated in a place the meter does not read: an edge, a seam, one side of a panel. The intervention is more measurement points and a visual check of the entire surface, not averaging one measurement. And if the cause lies in the tolerance, the difference falls within the agreed ΔE but the eye is more sensitive in that color region than that single threshold. The intervention is then revising the tolerance or supplementing it with limits on the individual components — lightness, chroma and hue separately.
The gain of this classification is that you shift the conversation from "who is right" to "which of the three is it". That makes the conflict solvable instead of a stalemate between two parties who are each right in their own way.
One source of the conflict deserves separate attention, because it is so often overlooked: the measurement geometry. An instrument measures at a fixed angle and in doing so decides whether or not to include the gloss. The eye moves, tilts the sample, and sees the color precisely through the gloss and the direction of the light. With glossy, metallic or textured surfaces, the two therefore structurally diverge: the meter can report a neat match while the eye sees a clear difference that lies entirely in the gloss and angle dependence.
That is not a measurement error but a difference in what is being measured. The lesson for practice is that with these kinds of materials you record the measurement geometry explicitly and use the same one on both sides of the chain, and that you perform the visual assessment with the same movement and lighting under which the product will eventually be seen. Where the meter and the eye disagree about glossy surfaces, the geometry is almost always the first thing you check — even before you doubt the assessor's judgment.
Do not treat your meter as the final judge and your assessors as mood-makers. Let the instrument quantify, let the trained eye judge what the number means in the application, and make sure that in a conflict you look for the cause instead of declaring one side "right". And make sure your eye has the authority it deserves — by knowing how well your people actually see.
If you want to get your entire assessment process in order — meter, light, reference and human — you can learn this systematically in the Color Assessment program.
Sources: Green-Armytage, P. (2006), The Value of Knowledge for Colour Design, Color Research & Application 31(4):253–269 (spectral profile vs. appearance, metamerism). — Hirschler, R. et al. (2018), How much colour science is not too much?, Color Research & Application 43:977–992 (metamerism, light source/observer). — Murphy, R. A. (2015), Comparing Color Vision Testing…, Pacific University (variation in color perception).
In a professional color process, 'red' is at least seven different things. Why that distinction is the foundation beneath reliable color management.
You agree on color with a supplier not with words but with L*a*b* and ΔE. This is how you record color, tolerance and measurement conditions unambiguously — and avoid surprises at delivery.
Eye and meter both assess color, but not in the same way. When to deploy a color meter, when the trained eye leads, and how to calibrate them against each other.