Colour theory and colour reproduction
Colour questions mix theory, measurement and control. You must know which model applies to which device, how a colour is measured and under which settings, and how a deviation is judged against a stated tolerance.
Models and spaces
- RGB is additive, used by screens and cameras; CMY(K) is subtractive, used by print. Lab is device-independent and is the reference for spot colours and tolerances.
- Hue, chroma and lightness describe a colour. Two samples can match under D50 and fail under D65; that is metamerism, and it appears when the pigments differ.
- Standard viewing for print is D50 with the 2-degree observer; displays are assessed at D65. Always state the illuminant alongside a visual judgement.
Measurement
- Densitometry gives density and tone value increase from a control strip, relative to the process; spectrophotometry gives spectral and Lab data, which is what delta E is computed from.
- ISO 13655 defines measurement conditions: M0, unfiltered tungsten; M1, D50 with UV included; M2, UV excluded. Papers with optical brighteners read differently under each, so never mix conditions in one report.
- Put total area coverage, grey balance and overprint patches on the control bar, not in the image area.
Proofing and deviation control
- A contract proof must represent the printing condition: ISO 12647-7 sets tolerances for the media wedge, and the proof must carry the OutputIntent and its measurement data.
- Spot colours are defined in Lab, not as CMYK builds; simulate them with fixed builds agreed in writing with the customer.
- Control deviation with statistical process control: track the mean and the 95th percentile of delta E across the run, not the single best sheet.
Worked example
Target L 56, a 10, b -36; measured L 55, a 12, b -38. Delta E 1976 = square root of (1 squared + 2 squared + 2 squared) = square root of 9 = 3.0. Against a tolerance of delta E 2000 of 2 or less, this fails; recomputed in delta E 2000 the value falls to roughly 2.1 to 2.3, still at the limit. Both a and b drifted negative, so the printed colour is too green-blue and slightly dark, which points at ink film or a batch change rather than at the paper.
Traps that cost marks
- Approving colour on an uncalibrated monitor in an office lit by windows.
- Comparing M0 measurements against M1 reference data.
- Quoting delta E 1976 when the contract specifies delta E 2000; the numbers are not interchangeable.