Threading the needle of colour

Colour science may have evolved, but modern set-ups are not immune to old challenges, discovers Phil Rhodes

Modern cinematography offers people a lot of options. That can sometimes be a euphemism for too many standards, and while controlling how colours render on cinema or TV screens has never been trivial, the interaction between all those options can make colour feel more technical and less creative than we’d like.

There was once a time when electronic cameras made pictures from one shade of red, green and blue. Brightness was defined by whatever a cathode ray tube could achieve. Back then, 700 pixels per line was a lot, and we were happy to accept that green simply meant whatever the phosphors in that tube would create. Even so, in the early eighties, long before electronic cameras became digital cameras, Panavision had been willing to adjust the internals of a standard-definition video camera (a CEI 310, which subsequently became Panacam) to make it more useful in drama production.

Maximum flexibility

Panacam did not take the world by storm, but it did begin a quest for better colour in electronic camera equipment, which is still piling options on us today.

Maybe a decade later, monitors started learning to make light using something other than glow-worm chemistry. Both brightness and colour range improved. Shortly after, cameras started to match that ambition with improved capabilities of their own, to the point that monitors had to start accounting for what cameras were doing. For decades, some of the most respected material ever shot used monochrome video taps. Now we happily select colour space, brightness encoding and a creative colour profile in pursuit of confidence that all those things are correctly configured.

It’s no surprise, then, that the modern demand for flexible, capable on-set monitoring soared at the same time as complexity in colour handling. The recent acquisition of Flanders Scientific by Atomos puts both the precision and affordable options under the same roof, consolidating what is now a mature market in monitoring that’s flexible enough to handle modern camera technique. Meanwhile, having added more moving parts to an already-complex machine, it’s unsurprising that DCS Labs’ line of precision colour reference materials (as well as more universal alternatives like the Macbeth chart) were widely adopted as a way of expressing what things should look like.

Later, trade show booths were stacked with monitors that were bright enough to test sunscreen, and with a deep enough primary green to handle mysterious concepts like deep turquoise, which had previously been beyond broadcast TV. Then, just as big streamers started demanding a wide colour gamut and high dynamic range, LED lights started offering colours to take advantage of it. Early LED designs gave us something similar to white light, but the temptation of full colour mixing, and the improved modern whites arising from more colour channels, was too much to resist. Recent technologies, typified by Aputure’s BLAIR engine, don’t even use a white-emitting LED to create quality daylight.

Today’s lights, cameras and monitors can handle almost any colour, and we haven’t even talked about post-production. Adobe’s recent interest in new colour features addresses the fact that people now walk into film school fully aware of how critical grading is, and it also emphasises that post-production software needs sophisticated colour handling to handle the variety of the job.

Consider also Kodak’s recent release of the VERITA 200D motion picture film stock 5206, which brings yet more colour options in the same way different stocks and different LUTs always did.

The reality, of course, is that colour has always been involved. The CIE 1931 diagram is five years short of its centenary, after all. It just took the flexibility of modern film and TV production equipment to make that part of an industry’s working reality.

Maximum consistency

The challenges aren’t new, and neither are potential solutions. Some of the earliest attempts to take the guesswork out of colour are showing their age. Characterising the quality of white light, for example, is such an old issue that CRI – a sixties solution – has itself become mostly obsolete in film and TV applications. If we insist on a simple higher-is-better number, more recent initiatives such as TLCI are certainly more useful, and the spectral similarity index (SSI) allows us to accurately compare two light sources for similarity.

Still, if anything embodies the tension between flexibility and consistency in the world of colour, it’s modern lighting control. The option to have any light be any colour at the tap of a phone screen is a time-saving, cost-saving revelation. The lost capability of having two lights from different manufacturers match at switch-on, conversely, is much mourned. Solutions are difficult, if only because manufacturing processes are variable enough that eyes and cameras can see the difference. To be fair, that was absolutely the case with HMI and fluorescent, but the sheer variability of LED has complicated the situation.

Within those limits, initiatives such as the ASC White Point, developed by the lighting committee of the society’s Motion Imaging Technology Council (MITC), have sought to define white light so that it can be implemented across devices and the things that control them. Ideally, this avoids the inconstancy of control mechanisms like hue, saturation and intensity, which usually don’t specify the colour that results from a particular set of numbers.

Sending CIE xy commands to lights is theoretically less equivocal, but crews typically prefer to discuss colour based on a correlated colour temperature (from reddish to blueish) and tint (from green to magenta). Those are familiar from filters and gels, even though the exact origins and definition of CCT and tint are not widely understood. Implementing plus- and minus-green features on colour-mixing LED lights involves some historical research on the nature of filters. The work of translating between CCT-plus-tint user inputs and CIE xy control data will often fall on lighting control devices, and not all (one exception being grandMA3) have implemented it.

Meanwhile, camera manufacturers have always seemed keen to keep colour and brightness handling proprietary, and as more cameras start to rely on post-production to handle large parts of the image processing pipeline, a universal solution seems unlikely. Old-school engineers might shake their heads at the lack of standardisation – or, again, the huge range of per-camera, per-light approaches, each designed around the priorities of a particular manufacturer and its engineering team.

The Reality

For sheer usability, the widespread adoption of a standard is unavoidably more important than that standard being ideal. Standardisation involves mutual acceptance of what something should do, inevitably circumscribed by what it shouldn’t. The early days of cinematography saw a huge variety of film gauges, many of which had to fall by the wayside before the world converged on a manageable number of options. It remains to be seen whether rationalisation will lead us back toward the halcyon days of tungsten-halogen lighting, which matched effortlessly, and without giving up all the conveniences that the modern world has developed.

There is some comfort, perhaps, in the knowledge that the human visual system is almost certainly the most sensitive colour-comparison instrument on any film set. If two sunbeams are well matched on a monitor, they will probably look well matched in the viewer’s lounge, too, even though studies remind us that humans can only remember colours for a few scant seconds. That may not have much to do with on-set arguments about the exact shade of a sunset.

Nevertheless, so long as all the sources of sunlight in a scene are the same colour, the likelihood of anyone continuing to complain that the sun is microscopically too green falls quickly after less than ten seconds of looking in the other direction.

This article appears in the June/July 2026 issue of Definition