8

What colors is an RGB LED actually able to emit? Obviously, it's not able to represent the whole spectrum of RGB (it will not be black, or brown).

I'm referring to a LED like this:

enter image description here

I found the following characteristics at https://www.analog.com/en/technical-articles/control-color-of-led-stage-and-architectural-lighting-easy-accurate-13-bit-color.html:

enter image description here

Is it true that a single LED is able to emit colors as the smallest triangle on the illustration above shows?

If this is the case, how does LED monitor display colors like brown?

mnj
  • 207
  • 2
  • 5
  • 5
    Look in the datasheet of the LED. *how does LED monitor display colors like brown?* Can the LED show brown? Who says that the smallest triangle represents the spectrum that the LED can produce? – Bimpelrekkie Oct 15 '21 at 10:08
  • 1
    I guess I should look at the wavelengths http://cdn.sparkfun.com/datasheets/Components/LED/YSL-R596AR3G4B5C-C10.pdf. I don't know if the smallest triangle represents LED colors range, it's my guess, and my question is whether my understanding is correct or not. – mnj Oct 15 '21 at 10:17
  • 1
    Black is the complete absence of light, you can't emit that. – Finbarr Oct 15 '21 at 10:38
  • 3
    Since they emit nearly monochromatic light at 460, 520 and 630 nm (typically, see datasheet for a specific product), they can produce a triangle spanning those points, which is nearly that entire diagram (much larger than the triangle). – user1850479 Oct 15 '21 at 14:51
  • 3
    Related: https://electronics.stackexchange.com/questions/437871/why-cant-rgb-or-bicolour-leds-produce-a-decent-yellow – winny Oct 15 '21 at 15:51
  • 7
    Regarding brown specifically, no light source on its own will ever look brown since brown is just dark orange. Similar to how you won't see a gray light, or a burgundy light. – Carmeister Oct 15 '21 at 19:59
  • A LED monitor is rarely consisting of individual LEDs for each pixel, it's a misleading term used by LCD manufacturers to refer to a LCD which is lighted by white LEDs. – Mark Ransom Oct 16 '21 at 03:17
  • If you want brown, try orange through a matte plastic filter. – Mast Oct 18 '21 at 17:44

7 Answers7

23

The simplest answer is… in the LED datasheet! Usually they give the emission spectrum of the single channels and from it you could work out the useful gamut of the led.

You might consider asking for binned parts if you really need consistence in your device emission.

Lorenzo Marcantonio
  • 8,231
  • 7
  • 28
22

An RGB LED cannot emit light at the yellow wavelength (around 590 nm). But it can emit a mixture of green and red light which looks exactly the same as yellow light to humans.

This is because our eyes do not sample the entire spectrum. We are trichromats and our three types of colour photoreceptors sample around the wavelengths for red, green and blue. 590 nm yellow activates both our green and our red receptors in a certain proportion. Any mixture of light that activates them in the same proportion will look the same as monochromatic, 590 nm yellow to us.

Interestingly, other animals may not agree at all, and some may be able to perceive the "RGB led yellow" as both red and green superimposed, without any yellow in it.

DamienD
  • 3,093
  • 1
  • 14
  • 23
  • 14
    _"We are trichromats"_ — [about 91% of us](https://en.wikipedia.org/wiki/Color_deficiency). Most of the rest are dichromats, while a tiny portion [is tetrachromatic](https://en.wikipedia.org/wiki/Tetrachromacy#Humans). As for _"590 nm yellow"_, it's actually orange. Really yellow, ranunculus-like, is about 571 nmD. – Ruslan Oct 15 '21 at 20:06
  • 13
    It may be worth noting that while the human eye which is viewing light directly might not be able to distinguish a mixture of red and green light from a monochromatic yellow light with a little blue mixed in, colored objects may appear different when viewed in such lights. Most objects that appears orange under white light would likely appear orange under a mixture of red and green light, but appear light or dark yellow when viewed with nearly monochromatic yellow light. – supercat Oct 15 '21 at 20:28
  • 8
    "our three types of colour photoreceptors sample around the wavelengths for red, green and blue" isn't really true... the red, green, and blue cones each sample from a wide spectrum; red cones respond more strongly to yellow light than red light, and green is far from the peak of green cones. Color perception is complex, but involves comparison between different cones. https://biology.stackexchange.com/questions/56385/how-are-colors-outside-the-standard-rgb-color-scheme-perceived for reference. – Bryan Krause Oct 15 '21 at 20:34
  • RGB LEDs can most certainly emit at 590nm. Look at the datasheets. Some do more than others, but these aren't lasers - they're quite broadband emitters. – J... Oct 18 '21 at 14:32
  • 1
    @Ruslan very few people are true dichromats (2%). The bulk of those who fail the Ishihara tests are anomalous trichromats -- still three receptor types, but two of them (usually red and green) are closer together on the spectrum, causing a different perception of polychromatic light and impaired discrimination due to the greater overlap. As one of these people I welcome your correction on yellow vs orange :p – DamienD Oct 19 '21 at 08:34
  • Hmm, so do an orange and a lemon look very similar to you? – Ruslan Oct 19 '21 at 08:36
  • 1
    @Ruslan depends. If they're very close together I can tell which one is more red or more green without any problem. Far apart, much harder. I use texture clues a lot and sometimes attribute colour casts to neutral grey on that basis. – DamienD Oct 19 '21 at 08:42
15

LEDs (pure color types without phosphor, which eliminates white, pink and some other odd colors) are almost monochromatic with some thermal broadening, so they emit light that is near the spectral locus of the CIE chromaticity diagram which you have reproduced above (the oblong shape with wavelengths listed for various points). The green color is not as close to the perimeter because of the curvature of the spectral locus in the vicinity of the center of the green LED wavelength distribution.

The gamut of reproducible colors of an RGB LED is a triangle joining the three LED colors, like this (the outer dashed triangle-- image from Fig 19.2 LIGHT-EMITTING DIODES, 2nd Edition, Cambridge Press):

enter image description here

The sRGB colors represent the color CRT phosphor colors of the sRGB color standard.

Brown in sRGB color space might be 58.8% Red, 29.4% Green, 0% Blue, or something like this (image from here:

enter image description here

Human perceived color, converting between color spaces such as the two described above, and so on, is a fairly complex subject. There are good books and some good info on the web such as this site.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • 2
    A good answer! Just FYI the correct name for the curve that you are referring to as the "outer perimeter" is the "spectral locus". The perimeter of the chromaticity diagram is the whole square from (0,0) to (1,1). – Tom V Oct 15 '21 at 18:48
  • 1
    @TomV Thanks! Edited for the correct nomenclature. – Spehro Pefhany Oct 16 '21 at 01:43
  • It took me forever to realize that brown is simply a dark orange. Nice illustration. – Mark Ransom Oct 16 '21 at 03:21
  • 1
    Note that 58.8% red, 29.4% green, 0% blue in your example are not percentages of _intensity_. Instead these are sRGB values, which are related to intensities via the sRGB transfer function (often approximated by a gamma of 2.2). – Ruslan Oct 16 '21 at 07:49
  • @Ruslan you're correct that the sRGB numbers are not linear measures of intensity. But they are pretty close indicators of apparent brightness, since the eye isn't linear either - it has a response curve approximating a gamma of 2.4. – Mark Ransom Oct 17 '21 at 04:20
  • @MarkRansom we are talking of electrical engineering here. What's important now is correct reproduction of luminance (and indirectly, chromaticity), in this case by applying appropriate currents to the LED subpixels. How humans perceive it is secondary; anyway, such a simple description as via any kind of fixed gamma is useless: there's [Weber-Fechner law](https://en.wikipedia.org/wiki/Weber–Fechner_law) that operates in terms of logarithms, then [Steven's power law](https://en.wikipedia.org/wiki/Stevens%27s_power_law) with _different_ exponents for visual stimuli, … – Ruslan Oct 17 '21 at 20:23
  • @Ruslan it's probably a mistake to assume it's being driven by sRGB numbers anyway, unless there's a circuit to apply the non-linear conversion. – Mark Ransom Oct 17 '21 at 22:24
  • @MarkRansom well, I suppose it'd be done by PWM from a microcontroller that does the necessary maths. – Ruslan Oct 17 '21 at 23:09
12

Brown is simply dark yellow or dark orange, and looks brown when juxtaposed with brighter colours.

I recommend this YouTube video. However, it just takes 21 minutes to say what my first paragraph says, with some illustrations and humour.

Neil_UK
  • 158,152
  • 3
  • 173
  • 387
  • and for those who'd like to descend even further down the color rabbit hole: [Billmeyer and Saltzman's Principles of Color Technology](https://www.wiley.com/en-us/Billmeyer+and+Saltzman%27s+Principles+of+Color+Technology%2C+4th+Edition-p-9781119366683) – uhoh Oct 17 '21 at 01:27
  • I'd add a summary of that video: "Brown really just is dark orange. What we humans perceive as brown is the wavelength of orange in certain contexts." – orithena Oct 18 '21 at 12:37
7

You're misinterpreting that graphic

That colour space is showing hues of equal saturation and brightness. Brown is merely one of those hues with lower brightness.

In colours, RGB is certainly how light emitters and receptors work. However it's a bad representation of how we perceive colours after the brain has done its image processing. HSV or HSL are far better "natural" representations - and as you'll see from the Wikipedia graphic, all the hues tend to fade through brown on their way to black (zero "value" or "lightness").

And RGB sometimes isn't good enough

LED stage lights have been around for a while. The appeal is clear from a lighting point of view, because you can sweep between colours easily and instantly. And from an installation point of view, power consumption means you don't need kilowatts of power distribution, and you don't need lots of expensive faders (which are always electrically noisy and screw with your microphones). And it's better for performers, because those kilowatts mostly went into heating the stage up.

But after people had started using them, they found some obvious flaws. RGB LEDs aren't perfectly balanced, so it's hard to get a good clear white - and more than that, it's hard to accurately get a tinted white light. The tint is important, because a blue cast gives you the effect of moonlight, a yellow cast gives you sunlight, and orange gives you sunset. Following on from that too, it's also really hard to get good accurate shades of yellow from RGBs, and that's important because yellow is what gives us "warm" or "cold" light. Human eyes have high resolution to hues between red and green, so we can easily notice when something isn't quite right there.

So better stage lights aren't just RGB - they're RGBWA, with additional white and amber LEDs. That lets the lighting technician set the "lightness" in HSL space (between fully white and fully black) much more accurately with the white channel, and the amber channel gives much better control of the hue and saturation for anything between red and yellow.

LED TVs have picked up on this too. It's taken longer of course, because it's really technically difficult to make an LED TV screen, and just getting decent yields from RGB was hard enough to start with! So far I'm not aware of any which use full RGBWA, but RGBW screens are certainly available and advertise sharper whites and crisper colours.

Graham
  • 6,020
  • 13
  • 20
  • 2
    Equal brightness, but not saturation. The saturation of white at the center is zero. – fraxinus Oct 16 '21 at 10:04
  • 3
    The bigger reason RGB is not used to produce white illumination is that it has unusably low CRI. Calibration is also annoying since the individual diodes drift with respect to each other, but even perfectly calibrated you cannot produce high CRI white from an RGB LED source. – user1850479 Oct 16 '21 at 17:10
  • RGB lighting is also incredibly irritating for glasses wearers because chromatic aberration in the lenses causes the red and blue point sources to separate out. If I look at a RGB white source off centre I see a blue LED and a red LED – Whelkaholism Oct 18 '21 at 15:13
4

You should consider the behaviour of the human eye when "different" subsections of waves come into the eye.

Think about magenta: what RGB LED emits is some amount of red and blue. But the brain mixes them and the final image in brain is magenta.

For your question, you are thinking about only one pixel. In the final image displayed on the TV, what you are seeing is not the colour of a tiny pixel. Instead, it's a mixture of the colours coming from adjacent pixels.

So, for brown, a single LED does not generate brown by mixing red and green with some amounts. Instead, adjacent pixels generate different colours to "approach" to brown so that your brain can mix them together for a "browner" image.

Rohat Kılıç
  • 26,954
  • 3
  • 25
  • 67
4

If this is the case, how does LED monitor display colors like brown?

It's interesting to note that this isn't how LED monitors work, but actually it kind of is.

"LED" monitors do not use RGB LEDs, they use white LEDs (which are themselves actually blue LEDs coated with a yellow phosphor) to backlight an LCD panel. However the LCD panel has red, green, and blue filters so all the answers about color mixing still apply.

OLED displays DO have individual RGB OLEDS for each pixel, as do microLED displays.

IceGlasses
  • 341
  • 1
  • 5
  • 1
    That's a very good point, see https://en.wikipedia.org/wiki/LED_display and the disambiguation paragraph there. – Carsten S Oct 18 '21 at 09:28