The Spectral Inadequacy of RGBW

One of the first¹ moving-head LED lights that I saw was the GLP Impression X90. Having spent the – well, not years exactly, but enough time in the biz to figure out that the MAC 2000 was more or less How We Did Things, it was pretty nifty. It was small, it was – in deep colors – punchier than moving heads three times its size, and it was blazingly fast in moves. It was a very cool light; a step in the ladder of advancement that had been in motion since we stopped burning bits of rock.

The Impression – and back then, we all just called it The Impression, because there weren’t others – made a real splash. Rarely had I seen a red that red out of a fixture². Such a sheen inevitably wears off after a while, of course, especially once you really dig into the colors. A certain sort of white came out of that front array of Skittles-like RGB emitters – a magenta-tinged mess that, if you twiddled the knobs hard enough, you could sort of browbeat into an approximation of an arc source. Dialing up anything south of 5,600K, though, and you were greeted with too red, or too green, but never quite the pale amber that you were going for. Trying to light a human with this would summon abbysal hair and skin tones from the depths, and no Lee filter or gel picker would help you now, mortal. The entire yellow range, in particular, was a weak point from the very beginning of RGB.

Of course, this is a practical consequence of using very narrow-bandwidth emitters to approximate a full spectrum of light. “Approximate” isn’t even the correct word, because it implies a sort of emulation. That isn’t what RGB is doing, really. Sure, it’s stimulating the cone cells and producing constituent parts of the photopic experience of yellow or white or whatever, but there are broad contours that get lost when you’re using the spikiest spikes and discarding the gentle slopes of the rest of the colors. You lose what is essentially a very pleasing sort of reduced contrast.

Then along came the Vari-Lite VLX³, audaciously upping our color count to four. You probably all know this by now, but every white LED you’ve ever seen began its life as a blue LED with a phosphor coating over the top to convert most of that blue light into a white spectrum. This helped immensely with the pastel parts of the color range, and we could finally point LED-based lights at humans without completely regretting it. But this is an imperfect solution tacked onto an already imperfect solution. The white isn’t a magical Fill-In-The-Gaps color that prevents metameric failure, it’s a blunt across-the-board tool. Because of the blue pump, early attempts produced a spectrum dominated by a blue spike, and even with today’s advancements we still have almost non-existent reds and ambers. White really does nothing to address the gaps in the spectrum, it just smooths them over for when you’re trying to make pastels.

What Lime – and to a similar extent, Osram’s Mint – did was to change some of that for the better, along with amber. “Direct amber” LEDs have been around for a while, but in 2009 Philips Lumileds created a phosphor-converted amber LED with a 30-40% efficiency. Adding to that, Lime and Mint fall right into a highly-sensitive part of your eye’s photopic curve, so a properly-tuned array of this quintet of LEDs (red, green, blue, amber, and Lime / Mint) produce some of the best color mixes we have available to play pretend-tungsten in this post-tungsten world. Emulating this should-be-dead source, however, isn’t the only thing we want to do, though it does suck up a lot of the color-mixing air in the room. Theatrical productions drive a lot of this, an understandable built-in zeal for tradition that comes from being dependent on metal filaments for the better part of one and a half centuries.

But I am largely concerned with concert design, so the synthesis of raucous and saturated colors is of particular import to me. Oh yes, give me your colors between the colors and every shade of incandescent blackbody emulation you wish – and I will pay for the privilege. But I also want a gold plucked straight from a summer sunrise, and a cyan bluer than any toilet bowl cleaner, and a magenta that Deutsche Telekom would sue you over. RGBW, however, never felt particularly great at reproducing these colors. It felt this way because is isn’t – in between your big tri-color spikes are deserts of spectral scarcity, which even the most untrained eye can discern, though unconsciously, as lacking.

Much work has gone into amber, and cyan, and deep red – parts of the spectrum still poorly-served by RGBW. Magenta is the outlier here, as magenta is an extra-spectral color that has no corresponding “single” wavelength – it’s a combination of red and blue wavelengths. The hard work shows, too – with the Source4 Series 3 deep red and indigo and amber and cyan emitters, I can make a truly delicious grape or vibrant rose color, or a fiery orange or a briny sea foam. Notably absent from the list of secondaries? Yellow.

Yellow is a tricky one. It’s difficult to produce with LEDs from a technical standpoint – whatever alchemy LED chemistry engineers do to discern the various elements or isotopes that will make a particular wavelength has been somewhat unfruitful, I am told, for yellow. Not totally, mind you, the chemistry itself does exist, but mostly in lab form where such LEDs remain under-powered and sadly somewhat under-researched relative to other colors. A high-powered phosphor-converted version could (probably) be made at scale, but the demand to fill the other gaps in the spectrum has been more urgent the last decade or so, as we’ve been trying mostly to create superior color rendering LEDs for houses and cars headlights. We weirdos in the entertainment biz are really the only people interested in having saturated colors anyway; mostly, people just want vanilla light. This is frustrating – when you compare the yellow you can get out of any LED system to that which you get from a subtractive system – especially an old-school gelled incandescent – it becomes apparent that you are glimpsing this color as though through a veil; a mere shade of the glory it can encompass. RGB yellow looks pale, sickly, weak. Even today, if I needed to get a deep banana or sunset yellow out of an LED fixture, maybe I’d dial up a calibrated white and put a gel in front of it, extracting the color I want from a full spectrum rather than relying on the red and green emitters⁴. Because I need that yellow. I crave it. The irony of saying I’d prefer subtractive mixing out of a PAR for one color is not lost on me, being on the record as being feverishly anti-incandescent, and I still am. I can accept this – I don’t think the benefits of LED are outweighed by not having a great spectrum for one slice of the rainbow, but I do find it nettlesome.

Various signs do point to a gradual improvement of the circumstances⁵. A higher percentage of lights newly on the market are upping their die color count relative to past attempts. GLP’s Impression X5, despite its moniker being highly suggestive otherwise, serves up four-color RGBL mixing. Vari-Lite’s VL5LED gives us RGBALC in a tasty VL5-like housing – initial reviews are favorable – and even the necrotic Strand Lighting has some RGBALC fixtures kickin’ around. Clay-Paky has the K-Eye, also with six colors. An important point bears raising here, and that is this: when I speak of the need for expanded color range fixtures, I’m speaking of a need that I see for a specific class of fixtures: moving head hard-edged and wash fixtures intended either for pointing at people, or creating big beamy looks in the air with haze, or painting color onto scenery or objects. Sometimes, a reduced spectrum like RGBW really is good enough.

I feel strongly that these expanded color range fixtures represent the way forward for the concert lighting industry. Undoubtedly, it will take time to totally divest ourselves of RGB(W) – there’s enough bric-a-brac rattling around in rental houses around the world that will keep powering up for years⁶ – but there’s always a need for low-cost wall washers or audience lighting or whatever application doesn’t require a spectral confetti cannon. I’m just wholly underwhelmed by the aggressive mediocrity of the RGBW fixtures I see these days. It’s not even the fixtures per se. Lots of companies are making these lights, and engineering them well enough – they’re solid workhorse lights I guess, and I don’t really have anything bad to say about them being what they are for the purposes they’re intended to serve. A spectrum can be good enough for the budget and use case scenario, and that’s whatever. But I want to see more innovation with our color systems and really move away from RGBW as the “default okay” color system that gets specified on shows.

I think the immediate response this might engender will take the form of a “diminishing returns” argument. I’m not wholly unsympathetic to that – rental houses in particular aren’t going to be crazy about buying six or seven or eight-color systems only a few years after investing in the new hotness of RGBW, and just to get to something that your average audience knucklehead won’t notice or care about. But I’m with Mike Wood on this one in believing that additive mixing is the way forward, and while I didn’t harp on this too much, I think people do notice, and excellence is its own reward⁷. To gain widespread adoption, we’ll have to fix these spectral inadequacies, of which yellow represents one of the Final Frontiers to be conquered. We’ll also have to figure out how to get more light out of fewer emitters while keeping things cool – no small feat there, either. You know, engineering.

My thesis: RGB is laughable at this point, RGBW only slightly less so, and RGBAL is the minimum needed for concert-grade color, and preferably RGBALC. Even that doesn’t fully cover all of what I’d like, but I’m unsure of the licensing / patent stuff that comes along with Indigo and Deep Red. (Why is only ETC using them? I dunno.) But even if we got to RGBALCIDe⁸⁹ systems, we’d still be missing a crucial band in the yellows. If there’s a point to all this, it’s a plea for continued research into closing the yellow gap with some high-power phosphor (or direct¹⁰!) yellow LEDs to fill in the spectrum, so we can finally experience the full color range with all of the advantages of LEDs. Perhaps through the time-honored tradition of collective outrage, we can “encourage” manufacturers into giving us nine or ten-color color-mixing systems – but only if we get some intelligent algorithmic software to go along with it. That, however, is an article for another time. In the meanwhile, when I need deep yellow, subtractive will be the way to go.