Colour Calibration: Accuracy Is Only Part of the Story

In any colour-critical environment, there is a deceptively simple question that needs to be answered: 

Whether you are grading a feature film, reviewing broadcast content, retouching a photograph or making any other decision where colour matters, the monitor is the point at which digital information becomes something we actually see. 

That makes display calibration fundamental to the workflow. But calibration itself is sometimes oversimplified. Connecting a measurement device, running some software and receiving a successful result does not necessarily tell us everything we need to know. 

We have to think about how the display has been calibrated, where those corrections have been made, how comprehensively its behaviour has been measured and, importantly, how we verify that performance afterwards. 

When calibrating a professional monitor, the most obvious place to start is usually the software supplied by the display manufacturer.

There is a good reason for this.

On monitors designed for hardware calibration, software such as EIZO ColorNavigator communicates directly with the electronics within the display and adjusts its internal Look-Up Tables, or LUTs.

This distinction matters.

An alternative approach is to modify the signal being sent from the computer’s graphics system. This can produce a calibrated result, but the correction is being applied before the signal reaches the monitor. Depending on the workflow and implementation, manipulating the graphics output can reduce the number of available output levels and potentially introduce unwanted quantisation or banding.

With hardware calibration, the objective is different: preserve the integrity of the source signal and perform the necessary correction within the monitor itself.

For professional applications, that is a significant advantage.

It also becomes particularly important in moving-image workflows. A professional reference monitor may receive its signal via SDI from dedicated video hardware rather than through a conventional computer graphics pipeline. In that environment, relying on operating-system or graphics-card calibration may simply not be appropriate.

It would be tempting to assume that if two monitors both offer hardware calibration, calibrating them is essentially the same process.

It isn’t.

The quality of a calibration is influenced not only by the monitor and measurement instrument, but also by the methodology used by the software.

Different manufacturers take different approaches to profiling their displays, generating corrections and validating the result. The number and distribution of colour patches measured can differ, as can the algorithms used to calculate the resulting corrections.

This doesn’t automatically make one approach right and another wrong. Manufacturer-developed calibration software can be extremely effective because it has been designed around detailed knowledge of a particular display’s characteristics and internal electronics.

But it does mean that we should avoid treating “calibrated” as a universal description of performance.

What matters is how that result was achieved and, ultimately, how closely the display performs against the required standard.

There is another consideration that becomes increasingly important as an organisation grows.

Imagine a facility operating displays from three or four different manufacturers.

Each manufacturer may provide its own calibration application. Each may have a different interface, workflow, terminology, networked management and recommended methodology. Staff need to understand all of them, while the organisation needs to establish whether the results produced by those different processes are genuinely comparable.

This is where specialist third-party calibration platforms such as Light Illusion’s ColourSpace and Portrait Displays’ Calman become particularly interesting.

Where supported by the monitor manufacturer, these platforms can communicate with the display and write calibration data to its internal calibration system rather than simply applying a correction at the graphics card.

The difference is important: organisations can retain the advantages of hardware-level display calibration while introducing a more consistent calibration environment across supported displays from different manufacturers.

For larger post-production, broadcast and colour-critical facilities, that consistency can be extremely valuable.

Third-party calibration systems can also allow users to build considerably more detailed profiles of display behaviour.

Rather than thinking about calibration simply as adjusting white point, luminance and a handful of primary and secondary colours, sophisticated profiling can measure a much larger volume of the display’s colour space.

That additional data can provide the calibration system with a more detailed picture of how the monitor behaves.

More patches do not automatically guarantee a better calibration. Measurement quality, patch selection, display stability, meter accuracy, probe profiling, signal path and the mathematics used to build the correction are all important.

But for demanding applications, the ability to interrogate the display in much greater depth can reveal behaviour that a simpler calibration routine might not identify.

This is particularly relevant when creating 3D LUT calibrations, where we are interested not only in greyscale but in how colours interact throughout the display’s colour volume.

The question therefore shouldn’t simply be, “How many patches did we measure?”

A better question is:

“Do we have enough information to understand the behaviour of this display to the level our application requires?”

There is another aspect of the process that I believe deserves considerably more attention: verification.

Calibration adjusts the display.

Verification tells us whether the result is correct.

Those are two different things.

A robust workflow should therefore include measurements before and after calibration. Specialist calibration systems can compare the measured performance of the display against a defined target and provide objective information about how closely it conforms.

That changes calibration from an act of faith into a measurable process.

And verification becomes even more valuable when those measurements are retained over time.

Displays change. The rate and nature of that change will depend on the display technology, operating hours, operating conditions and many other factors. Regular measurement creates a history of the display’s performance.

Instead of simply asking whether a monitor passed its calibration today, we can begin asking more useful questions.

How has its performance changed since the previous calibration? Is that change consistent with what we would expect? Are particular areas of performance moving faster than others? And, eventually, is the monitor still suitable for the role it is being asked to perform?

That turns calibration data into something much more powerful: evidence about the health and consistency of the display estate.

This becomes particularly important when we stop thinking about individual monitors and start thinking about entire workflows.

A colourist may have a reference monitor in a grading suite. Another display may be used for QC. There may be client viewing displays elsewhere in the facility. Content may subsequently be reviewed by another team in another building or even another country.

The real objective isn’t to have a collection of individually calibrated monitors.

It is to create a visual environment in which decisions can be made consistently.

That requires appropriate displays, controlled viewing conditions, reliable measurement equipment, a defined calibration methodology and ongoing verification.

For organisations operating mixed display estates, third-party platforms can provide an important common layer across that process. For organisations standardised around a particular professional monitor platform, manufacturer software may provide an extremely efficient and tightly integrated workflow.

Neither removes the need to understand what is actually happening during calibration.

As display technology develops, our approach to calibration needs to develop with it.

HDR, wide colour gamut displays and increasingly sophisticated OLED and dual layer LCD technologies are asking monitors to reproduce a much greater range of luminance and colour than traditional SDR workflows. At the same time, production environments are becoming more interconnected, and content is being viewed across an increasingly diverse range of displays.

As Dynamic range and brightness have increased, the tolerances involved in making critical visual decisions have not become less important. If anything (remove this), they have become more important.

That is why I believe we need to move the conversation beyond simply asking whether a monitor has been calibrated.

We should be asking  how it was calibrated, what was measured, where the correction was applied, how the result was verified and how that performance is changing over time.

The purpose of calibration isn’t to make a monitor look good.

It is to establish a known, repeatable relationship between the signal and what we see.

Because when someone makes a creative or technical decision based on an image, the most important thing a monitor can provide isn’t impressive colour;  it’s confidence that the colour can be trusted.