Print Colour Accurately

Accuracy is not a property of expensive equipment. A desktop inkjet, profiled properly and running consistently, will beat an industrial press that nobody has characterised. The method matters more than the machine.

Before anything else: consistency

A device that will not repeat itself cannot be corrected, because there is nothing stable to correct. Two things to establish first:

Across a sheetThe same colour printed at the top and the bottom should measure the same. Ink laydown, heat and pressure all vary across a press.
Between sheetsThe tenth sheet should match the first, and the thousandth should match the tenth.

Inkjet is usually more consistent than laser, and both are more consistent warm than cold. If either check fails, stop — profiling an unstable device produces a profile of one particular afternoon.

Step one: an ICC profile

A profile describes what a device does with a specific set of parameters: resolution, speed, screening, ink limit and substrate. Change any of them and the profile no longer describes what is happening.

So record the settings alongside the profile. A profile with no record of the conditions it was made under is a profile nobody can trust six months later.

Profile source
Your own, measuredThe right answer. Most measurement instruments come with profiling software; ChromaChecker also builds them.
From the paper or printer makerA reasonable start, and made on their machine rather than yours. Fine for undemanding work.
A reference profileDescribes a standard printing condition, not your device. Useful for checking what a colour should do, not what yours does.

The profile also tells you the device gamut — which colours are reachable at all. Ranges answers the same question against nineteen reference conditions, before a colour is chosen.

Step two: iterative correction

A profile is a general description of a device. It is fitted across the whole colour space, which means it is a compromise everywhere and exact nowhere — and for a handful of brand colours that everyone will scrutinise, a compromise is not what you want.

The method is straightforward and it works:

  1. Print the target colours through the profile.
  2. Measure what actually came out.
  3. Compute the correction for each — a small adjustment to the values sent to the device.
  4. Print again. Measure again. Repeat until the difference stops shrinking.

Two or three rounds normally take a handful of critical colours from “acceptable” to “indistinguishable”. The corrections are stored as a lookup table and applied on every subsequent job, so the work is done once per device and substrate.

ChromaChecker’s Color Match does this. It needs a printer, a supported instrument connected to CC Capture, and a subscription.

What correction cannot do. Move a colour into a gamut it is outside. If the device cannot reach it, iterating will converge on the closest thing it can print and stop there — which is the correct behaviour, and a reason to check Ranges before the colour is chosen rather than after.

Then check it under the right light

A print that measures correctly under D50 and is judged under a shop fluorescent has not been checked. Which condition to view under depends on how it was measured — measure on M2, view under P3 or P4; measure on M1, view under P1 or P2. See Measuring and Viewing.

And if the colour is in the blue-violet region, check it under more than one lamp before signing it off. Those are the colours that move most — Colour Inconstancy.

Read next