I’m very interested in LED lighting and the different spread of wavelengths that are produced compared to other types (like tungsten).

Given that a White LED bulb actually produces 3 different peaks in the Red, Green and Blue wavelengths. Why do we never experience artefacts of this in reality?

For example, suppose I put an Orange in a room and light it with a tungsten bulb. Let us suppose that it reflects only a very small band of wavelengths between 600 and 610 nm. The light bounces off the fruit and reaches my eye. The light stimulates (mainly) my Red and Green sensors and I perceive this colour as Orange.

However, if illuminated by R, G, B light with the Green at 530 nm and the Red at 680 nm, there will be NO light of the right wavelength available for reflection from the Orange and it will appear BLACK.

Yet we never see this effect, even to the slight degree that a colour would look “odd”. Is this purely because ALL colours around us are – by pure chance – very good at reflecting a wide band of colours AROUND their perceived colour which happens to include the RGB peaks – or is there some other effect taking place?

When an LED bulb has a frosted envelope enclosing it, does this have the effect of spreading the spectrum, or are the peaks still equally distinct after passing through the envelope? In other words, can the envelope shift the wavelength of some of the light on the way through?

  • 1
    $\begingroup$ The spectra of light sources do produce color artifacts. This is coarsely reflected in quantities like the "color rendering index" (cri), which attempt to give simple means of judging the technical color reproduction qualities of commercial light sources. You are correct, though, if we wanted to capture the entire range of effects, we would have to deal directly with the spectra, which is complicated. $\endgroup$
    – CuriousOne
    Jan 5 '16 at 17:24
  • $\begingroup$ I remember from my photography days that tungsten light is orange, flourescent is light green etc. but I still think of them as producing a fairly good SPREAD of colours albeit that they are deficient in certain wavelengths. Our brains compensate for the colour bias so well we don't notice. It just seems to me that LEDs are an extreme version of this phenomenon - and our brains can't see orange colour when the object is, literally, black. $\endgroup$
    – Lefty
    Jan 5 '16 at 17:57
  • $\begingroup$ I can't put it into photographic terms for you, but the concern is valid. One can clearly see the differences between early LED lights with CRIs in the low 60s and current designs with CRIs >90. The phosphors used in these lights are having wider spectra which begin to approximate thermal light sources, but there is still a blue/purple peak from the driving LED. I would expect to have almost ideal solid state lighting within the next two decades, or so. It is not an easy technological problem, but the physics is understood and the lighting industry is working on it. $\endgroup$
    – CuriousOne
    Jan 5 '16 at 18:09
  • $\begingroup$ Google for metamerism and metameric failure. $\endgroup$ Jan 5 '16 at 19:02
  • 1
    $\begingroup$ P.S.: Most white LEDs do not have a spectrum with red, blue, and green peaks. Most emit a spectrum with a narrow blue peak (from the LED itself), and a very broad, yellowish spectrum from the phosphor. en.wikipedia.org/wiki/Light-emitting_diode#Phosphor-based_LEDs $\endgroup$ Jan 5 '16 at 19:05

What you are talking about is referred to as the Color Rendering Index (CRI) of a light source. I discussed this in a few earlier answers - this, that and particularly this one

Just to recap: if the illumination spectrum does not match the spectrum of white light, a particular color may be perceived as something completely different by the brain. The trick to creating a "good" white light source is the mixing of the phosphors that achieve this effect. Simply using three quasi-monochromatic sources of light (red, green, blue LED) is a really poor way of achieving white illumination: although the eye might perceive the right mixture as white if you look directly at the light source, there's really no telling how you will perceive colors that fall outside of the spectra of the illumination.

And that's precisely what the CRI measures.


First of all, White LEDs are usually NOT a combination of RGB lights. Most white LEDs are the combination of a blue LED with a yellowish fluorescent coating that transforms part of the blue light into yellow. The color-temperature of White LEDs is tuned by changing quantity/type of this fluorescent material.

This is how the spectrum of a 5000K white LED looks like:

nichia white LED 5000K

It is a Blue LED with a peak frequency around 450 nm, coupled with a fluorescent material that converts the blue light into a large set of other wavelengths between roughly 450 & 750 nm, with a peak frequency around 560 nm.

This should answer your first question: the obvious artefacts you are talking about do not appear because even though it is mountain-shaped, the white LED spectrum is continuous: all wavelengths of the visible spectrum are present.


Now, about the orange. The reflectance spectrum of the orange is also continuous:

Real spectrum of the skin of an orange

There is no reason the orange will appear black as you say, even with RGB white light, as the light will still be reflected by the skin of the orange.

However, actually you had a good intuition since, as the comments below already point it out, the artefacts exist and the RGB or standard LED white do not render the same colors as, say, the Sunlight.

Notice that if materials could be tuned to reflect 100% of some selected wavelengths narrow bands, these materials would be very colorful and very dark at the same time, as they would absorb most incoming light.


About your third question the answer is NO. The diffusive quality of the envelope (transparent, frosted...) does not significantly changes the spectrum of the light.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.