What To Look For When Judging An LED Fixture's Colour Mixing Capabilities...

What To Look For When Judging An LED Fixture's Colour Mixing Capabilities...

Colour Mixing and LEDs

We’ve all seen examples of it. You buy an RGB LED fixture that claims 16.7 million colours. You plug it in... yet out of those 16.7 million available colours, you seem unable to find any of the subtle hues, deep blue or even white or amber.... In fact, any colour except red, green, blue, cyan or magenta. So why is this? What’s so different about LEDs that makes mixing them so much harder than most other RGB or CMY mixing systems? As always, there is a combination of factors at play. In this article, we’re going to examine some of the reasons behind this, with a focus on the theory and science behind colours and light, and examine how that affects the effectiveness of colour mixing.

What’s different about the output of an LED?

Well firstly, they’re a lot narrower. The diagram below (courtesy of ETC’s blog on their new Selador range) compares a red LED with a red gel (tungsten lamp):

RedvsGel

The diagram clearly shows a much gentler curve on the gel. It’s worth remembering the human eye can’t see much past 700nm, but even so, there’s a clear difference in the spectra behind the colour. 

Now, let’s compare this with how our eyes respond to colour. Those of you who remember your biology classes may recall that our eyes respond to light through cones and rods. The rods are good in low level light conditions, but appear to be inactive in normal lighting conditions. The cones, on the other hand, work well in normal lighting conditions. There are three different types of cones that each respond to different wavelengths of light, and it is through comparing the intensity of light detected in each type of cone that we discern colour. Rods, on the other hand, come in one type only, so there is nothing to compare against, which is why (when in low light) we only see in black and white. Since we can’t see colour in low light (when the rods are in use) we’re going to focus on the cones only. The graph below shows the sensitivity of the three types of cones to different wavelengths of light:

cones_web

What I find most interesting about this graph is how wide the reception of the cones is. We can see the red cones can pick up heaps of green, and peaks more around amber than red. We can also see the distribution between blue, green and red is not equal: the green curve lies much closer to red than it does blue. As an interesting aside, some creatures appear to have a fourth type of cone: one that lies predominantly in the UV range. Experiments have shown a wide variation of cone configurations between species, with a few animals having only one type, many two or three and a few all four.

How does this fit LED?

Let’s look at the red, green and blue LEDs on top of the diagram above showing the reception of our cones. The spectra for these LEDs were pulled from the Philips “Luxeon Rebel” datasheet.

luxeon_vs_cones_RGB_LED

The first thing you’ll probably notice is how narrow the spectrum of the LEDs are. This is significant as it’s clear our eyes are capable of seeing a much broader space. Not only that, but we can see there’s a lot of overlap between our eye’s green and red cones (making our eye particularly adept at noticing subtle changes in that area), but with the LEDs, there’s this almighty chasm between the green and red LEDs. It’s not just between green and red that’s a problem either: green-blue and blue-purple are both under-represented when using RGB LEDs. 

As a result of this, many LED fixtures are including colours beyond the three primaries. Let’s have another look at the graph above, but this time with all seven colours in the Luxeon Rebel range:

luxeon_vs_cones_ROAGCBI_LED

It’s easy to see how inserting extra colours can make a significant difference to the colour rendering abilities of a fixture. We have already seen this recognised in many fixtures out there. The Selador range, which uses the Luxeon Rebel LEDs above, uses all seven colours to fill in as many gaps as possible. Later on, we’ll look at some chromacity diagrams which’ll further demonstrate this point.

Before we examine the benefits of colour mixing with more than the primaries, let’s look at how 3- colour mixing works with white lamps gelled in L079 (just blue), L139 (primary green) and L106 (primary red) – the LEE colours often used for achieving primary red, green and blue:

luxeon_vs_cones_RGB_Gel

Hardly ideal, but better than three colour LED. The green will need a fair bit of boosting and, behind a tungsten source, so will the blue, but we’re starting to get some proper overlap, and the curves actually have some meat to them. Still, there’s a lot of empty space between the green and red again - right where our eyes are most sensitive. Just as with LED fixtures, lighting designers have often resorted to using an extra colour or two to fill in the gaps (it is common for cyclorama washes to use four colours: red, green, blue and amber).

Would the REAL red, green and blue please stand up?

So what can we learn from these graphs? Well, for starters, the terms primary red, green and blue are too vague. Both a “primary red” LED and a “primary red” gel look like pure red, but behind the scenes there’s a world of difference. This problem has been understood for quite some time in the world of graphic design – where one computer monitor might use completely different combinations of red, green and blue to achieve the same colour of another. The Commission Internationale de L’eclairage (International Commission on Illumination), or CIE has defined “ideal” red, green and blues – with which we can produce every colour the human eye can see, seen in the graph below:

luxeon_vs_cones_RGB_Ideal

What should be immediately obvious with these “ideal” curves is that parts of them fall below zero (especially the red) which, in the real world, is impossible. So, these theoretical “ideal” primaries cannot be achieved in the real world and hence, despite what you may have been taught by your over-zealous primary school teachers, you cannot achieve every colour the human eye can see with just red, green and blue. 

While it’s nice to know what red, green and blue is required to achieve all colours, it’s not terribly useful to those of us producing the photons. In the computing world, it’s been accepted that we can’t produce all colours, and in an effort to standardize between devices, have defined “colour spaces.” The most commonly used is sRGB, which was defined by HP and Microsoft (image from Wikipedia): 

542px-CIExy1931_sRGB

The coloured area is the full gamut of what the human eye can see. The area within the triangle is the subset of those colours that sRGB can reproduce (D65 is the “white point”). By defining a colour space such as sRGB, it is well known what constitutes red, green and blue, and devices can be factory set to keep their primary colours as close to those as possible, or can use colour profiles to translate the monitor’s gamut to one that resembles sRGB. High quality devices can produce colours well beyond the limited sRGB gamut so alternative gamuts exist, such as the Adobe RGB colour space. In conjunction with the colour profiles, images created or modified in one colour space can be translated to another one as closely as possible.

Colour Spaces in the Lighting World

Unfortunately, in the lighting world, there are no standard, well-defined colour spaces which fixtures can work within. It’s probably not practical anyway: LEDs change colour and lose intensity over time, MSRs do the same, and tungsten lamps change as they’re dimmed. As these variables change, the colour spaces would need to be updated, which isn’t practical. Even so, it’s still worth being aware that devices emitting light do it within a specific part of the spectrum. Your colour-mixing fixtures still have a colour space; it’s just not neatly defined in a colour profile that allows translation between fixtures with a different colour space.

The best thing we have to assist with colour matching between fixtures is to select fixtures that use matched LEDs. Because of the unavoidable variations in colour between supposedly identical LEDs, many LED manufacturers perform a bunch of tests on their LEDs and sort them into separate “bins” based on criteria such as intensity, colour and voltage. This is a process that has only begun recently and has resulted in much closer colour matching. If you want your fixtures to be the same colour from unit to unit, choosing a model that uses LEDs from the closest colour matched bins is important. It won’t help when using two fixtures from different manufacturers, but at least a row of the same type of fixture will all appear the same colour.

In an attempt to express the range of colours a fixture is capable of producing, an often quoted specification in lighting instruments is the colour rendering index, or “CRI.” This is (supposedly) a measure of how closely the fixture produces the full spectrum of light that makes up natural white light (and therefore the range of different colours available). The CRI is obtained by comparing the “white light” output of the fixture with the ideal waveform for white light. However, CRI can be misleading: it does not highlight if the white light on which it is based is at a suitable colour temperature. It also treats all colours (points on the spectrum) equally, when our eyes do not. For these reasons (and some others) CRI is useful as a guide, but you should always judge the colour mixing abilities of the fixture for yourself.

One last thing worth noticing in the sRGB colour profile we mentioned earlier is that the sRGB colour space is a neat triangle – and each point of the triangle is the location of the primaries. So, when mixing three colours, your colour space will be defined within three points. If mixing four colours, your colour space will be defined within four points and so on. Below (once again from ETC’s Selador blog) is a comparison between a fixture using only the red, green and blue Luxeon Rebel LEDs and one using all seven of them. 

gamuts_web

We can see the colour space in a standard 3-colour LED fixture on the left is already larger than the sRGB colour space we saw earlier, but still has large areas missing. The 7-colour LED fixture on the right fills in many of the weak points of the conventional 3-colour gamut, which reinforces what we saw earlier in the spectra diagrams.

Colour Shadows

Finally, we should talk briefly about the chromatic shift in LED fixtures. Most LED fixtures in the market place at the moment use a separate lens and reflector for each LED – which makes each red, green and blue LED distinguishable from one another, and does not mix them. If lighting a subject from a long range, these separate sources generally merge into one. However, if the source is visible, or the throw (distance between the lamp and subject) is not that long, the separate red, green and blue colours can be distinguished, either in the sources themselves, or in the shadows that surround a subject. If visible, this colour separation can have a significantly detrimental effect on the perceived colour mixing. Below are a couple of pictures. On the left is the original Chroma-Q Colour Block, which included separate lenses for each LED. On the right is the new Chroma-Q Colour Block 2, which, while having a red, green, blue and amber LEDs behind each of the four clusters, produces much better colour homogenisation in the various mixed colours presented in the picture: 

colourblock_web

As can be clearly seen, the Colour Block 1 (even though in open white) does not provide a clear sense of a single colour, and will not as long as the viewer can see the component colours. The Colour Block 2, on the other hand, does a much better job at mixing and providing a good range of colours (we can see four different mixes above) with no hint of their component red, green and blue.

Test your fixture

As with so many things in life, there is no single way to claim one fixture is “better” than another at colour mixing. We’ve examined a number of reasons why colour mixing doesn’t always work, and what techniques and strategies are being undertaken by manufacturers to combat these issues. In the end though, nothing beats testing the fixture yourself. Try to get hold of a demo unit and put it through its paces. 

  • Consider, will the fixture’s source be visible? Light a subject at a throw range similar to what you intend to use it for and pay attention to the colour mixing not only on the subject, but in the shadows surrounding them, too.
  • Consider if you need bright, saturated colours out of the fixture or muted pastels or tints? 
  • Consider some colours you like to use with your conventional fixtures now; can you achieve something close with your fixture, and does it still look good on the stage? 
  • Compare one unit next to another, and judge if its colour matching is suitable for your application.


There’s never a silver bullet to solving these issues, but at the very least, hopefully this article has shown you what to look for when judging LED colour mixing yourself.

Blake Garner
Lighting Applications Engineer – Jands Pty Ltd

Bookmark and Share

Comment

Please comment using the form below (email will not be displayed)

Name

Comment

Rating

Email

Please note that your comment will only appear after it has been moderated.

Captcha

Security key