Short answer
A batch is approved by comparing a prepared sample with the standard twice: by eye in a light booth under agreed sources, and by instrument against a numeric tolerance. The two are not duplicates. The instrument gives a repeatable number for one geometry and one illuminant; the eye judges the whole appearance, including gloss, texture and what happens when the light changes. A sound procedure therefore states who is qualified to look, under which lamps, and which check decides when they disagree.
Instrumental approval exists because eyes tire, differ and cannot be audited; visual approval exists because the customer will not be using a spectrophotometer. A reading is a number for one measuring geometry, one illuminant and observer, and one small area. A look takes in the whole piece at the angle and distance it will be used at, including surface effects no single reading captures. Factories that rely on the number alone ship batches that measure well and look wrong, and those that rely on the eye alone cannot explain a rejection or hold a second shift to the same judgement. The standard practice is to do both, and to treat a disagreement between them as information about the specification.
Visual appraisal is standardised so that the judgement can be repeated. ISO 3668 describes comparing paint films against a reference or a freshly prepared standard under artificial light in a booth. ASTM D1729 covers diffusely illuminated opaque materials generally and recommends simulated average daylight, D65, with D50 where colour photography or printing is involved; AATCC EP9 gives the equivalent procedure for textiles. All of them control the light, the surround and the arrangement of the specimens. They also control the person. ASTM D1729 asks for observers with at least normal and preferably superior colour vision, and a companion guide covers selecting and training them, on the ground that instruments approximate appearance without reproducing the interpretive step of seeing.
No lamp reproduces D65 or D50 exactly; a booth contains a daylight simulator, and simulators differ. The joint CIE and ISO method grades one by calculating how far pairs of virtual specimens, which match exactly under the true daylight illuminant, separate under the simulator. That is a metamerism index, chosen because metameric pairs are exactly the samples for which an imperfect lamp gives the wrong verdict. Two consequences follow for approval. A booth's grade and lamp hours belong in the procedure, not only its nominal colour temperature. And a second, deliberately different source is used to hunt for metamerism: a batch that matches in daylight and separates under the shop-lighting source has been matched with different colourants.
Print uses its own viewing standard: ISO 3664 specifies D50, with a bright condition for critical comparison and a dimmer one for practical appraisal.
The numeric tolerance decides most batches without argument, and the useful refinement is an action margin. A limit is not a cliff: a batch at 0.99 of the limit and one at 1.01 are visually the same, so software commonly marks a band just inside the limit in which a batch passes but is flagged. One instrument maker illustrates this with a ten per cent margin. Batches in the band are the ones to look at in the booth, and a production trend entering it is the signal to correct before anything fails. Reading the lightness, chroma and hue components alongside the total shows which way the batch has gone, which the single figure cannot.
Disagreement has a short list of causes, and each points somewhere specific. If the instrument passes and the eye fails, look for a hue shift inside a tolerance of the wrong shape, a gloss or texture difference, or metamerism under the second lamp. If the eye passes and the instrument fails, suspect sample presentation, a drifted working standard or a limit set tighter than customers have ever needed. ASTM's practice for calculating tolerances says outright that they should be correlated with visual appraisal. The procedure should therefore name which check prevails, usually the visual one within a stated instrumental outer limit, and require the disagreement to be recorded, because repeated disagreements mean the tolerance needs rebuilding.
Why: The visual check depends on untested observers, different lamps or daylight leaking into the booth area.
Fix: Test observers' colour vision, fix the booth source and surround in the procedure, and log lamp hours.
Why: The match is metameric and was only viewed under the daylight source.
Fix: Make the second-source comparison a mandatory step and add a metamerism limit to the specification.
Why: There is no margin inside the limit and no rule for which check prevails near the edge.
Fix: Define an action band, send only batches inside it to the booth, and record the visual verdict against the reading.
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
ISO 3668:2017 specifies visual comparison of the colour of paint films against a reference standard or a freshly prepared standard using artificial light in a standard booth, and does not apply to special-effect coatings unless all illuminating and viewing details are agreed beforehand.
Caveat: From the standard's public abstract; booth illuminance and panel sizes are in the paid text.
Source: ISO 3668:2017 Paints and varnishes — Visual comparison of colour of paints
StandardStrong evidence
ASTM D1729-22 recommends simulated average daylight, D65, for visual appraisal of diffusely illuminated opaque materials, recommends D50 for applications involving colour photography or colour printing, and calls for observers with at least normal and preferably superior colour vision.
StandardStrong evidence
The joint CIE and ISO method (CIE S 012 / ISO 23603) grades a daylight simulator by the special metamerism index for change of illuminant, using pairs of virtual specimens that are metameric matches under the CIE daylight illuminant for the 1964 standard observer.
StandardStrong evidence
ASTM D2244-25 states that colour tolerances should be correlated with visual appraisal carried out under Practice D1729, and that gloss, texture and how close the specimens are held affect how a measured difference relates to commercial acceptability.
ConventionModerate evidence
Quality-control software commonly applies an action margin inside the tolerance, so that a batch close to the limit passes but is flagged for visual evaluation; one instrument maker's worked example uses a 10 % margin on a limit of 1.0.
Caveat: The 10 % figure is an illustration in a manufacturer's white paper, not a recommended value.
Source: Tolerancing: The Key to Accurate Color (white paper L10-616)
StandardStrong evidence
ASTM E1499-16(2023) gives criteria for selecting, evaluating and training visual observers for colour, colour-difference, gloss and metamerism work and for setting tolerances.
Source: ASTM E1499-16(2023) Standard Guide for Selection, Evaluation, and Training of Observers
Reviewed 6 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.