The amazing camera Part 1: Color vision

The miracle of vision

The human eye is a camera with 6 million color-sensitive pixels (cones) and 120 million black-and-white pixels (rods). It has autofocus capability: it can change the shape of its lens to bring close or distant objects into focus. The retina has "wiring and circuitry" behind it that accentuates contrast or detects motion. The brain does a lot of processing to turn the retinal images into meaningful information about a changing three-dimensional environment. The seemingly simple act of catching a baseball relies on impressive feats of image sensing and processing. No wonder the word "vision" has metaphorical meanings relating to qualities such as foresight, discernment, and wisdom.

But our eyes can also fool us. It's important to understand the limitations of naked-eye observations and why doing serious science requires instruments to go beyond those limitations. In this post we'll look at color perception. In a follow-up post we'll discuss perspective, depth perception, and visual acuity.

A tiny sliver of the spectrum

What we call "light" consists of waves of electric and magnetic fields. The wavelength (the distance between peaks or valleys of the wave) lies on a spectrum. Here is a picture showing this spectrum:

The electromagnetic spectrum. Above: objects the same size as the wavelength. Below: the visible range is expanded to show the wavelength ranges of the three types of cones in the human eye.

At the long-wavelength (left) end of the spectrum, there are very-low-frequency waves such as those you'd pick up from 60-cycle-per second power lines. Walking across the figure, the wavelengths get shorter and we encounter radio waves, microwaves, infrared (IR) light, visible light, ultraviolet (UV) light, and then x-rays and high-energy radiation such as gamma rays. Our eyes are sensitive only to visible wavelengths, a very narrow range of wavelengths close to the size of a small bacterium.

Images of Centaurus A in different regions of the spectrum

If we had to rely on naked-eye observations alone, the progress of science would be extremely limited. Above are four images of the galaxy Centaurus A taken with imaging instruments sensitive at different wavelengths. The images look different because different wavelengths are produced by different physical processes. The dusty core of the galaxy is best seen in visible light, infrared light shows up the temperature of many objects, and high-energy jets give off x-rays.

What is color?

The retina has three different types of cones, sometimes called R, G, and B for red, green, and blue. But, as shown in the above spectrum figure, each type of cone is sensitive to a range of wavelengths, and their ranges overlap. The R cone, for example, has peak sensitivity at a wavelength most of us would perceive as yellow, and its sensitivity tapers off as the wavelength gets either shorter or longer. A source of light will have a mixture of many different wavelengths stimulating the three types of cones and producing a different response from each type. The brain responds to the relative strengths of the three responses to produce the sensation of color. So color should not be confused with wavelength. Wavelength is a physical property of a light wave. Color is a neurological response to a wavelength or to a mixture of wavelengths. You can think of color as two ratios: the strength of the R cone's response relative to the B cone's and the strength of the G cone's response relative to the B cone's.

Let's illustrate with an example of two different light sources that produce the same color.

Three lasers with different wavelengths. We can mix the red and green lasers to produce the same color as the yellow laser.

A laser is a light source with a single wavelength (or very close to it). Suppose we have three lasers with respective wavelengths of 546 nm, 570 nm, and 700 nm. (nm or nanometer stands for one billionth of a meter.) These will appear respectively as green, yellow, and red to the human eye. If we project the red and green lasers to the same point on a screen, the eye will see a yellow spot. We can adjust the brightness of the red and green lasers so that this spot is indistinguishable color-wise from a spot made by the yellow laser. The two spots have different spectra (wavelength combinations), but they stimulate the R, G, and B cones in the same ratio. In fact, there are many different spectra that will produce the same yellow; these spectra have physically different wavelength mixtures, but their colors will be indistinguishable to the naked eye. This tells us that the eye is capturing only very limited information about the spectrum of any light source it sees.

We can also combine wavelengths to get whites or off-whites or other colors that don't match any single wavelength. Here's an example relating to stellar astronomy.

Combining red, green, and blue wavelengths to match the color of a star

The diagram on the left above shows the colors that we can get by combining wavelengths. The periphery of the region shows the wavelengths of lasers that we might imagine buying and how each would appear in color. For example, a laser with a wavelength of 520 nm is the point at the very top; it would look green. The interior of the region shows colors we can get by combining light from multiple lasers. Suppose, for example, we combine red, green, and blue lasers at the wavelengths shown on the right. We can adjust the brightnesses of the lasers to match the color of the sun; from space that's close to a white color. If we crank up the brightness of the blue laser and crank down the brightness of the red laser, we can match the color of Rigel, a very hot star with a bluish color. If we adjust the lasers the opposite way, we can match the color of Antares, a cooler reddish star.

Spectrum of the sun (without losses from the atmosphere)

Naked eyes are not enough!

As shown above, the spectrum of a star contains a continuum of wavelengths ranging from the ultraviolet to well out into the infrared. The spectrum of stars is a cosmic thermometer; the hotter the star, the shorter the wavelength where the spectrum peaks. Rigel is blue and Antares is red because Rigel is much hotter. While we can get a rough idea of the star's temperature from observing its color, precise temperature measurements require more detailed information about the spectrum. Having only three types of cones doesn't get us there.

The spectrum of stars is also a cosmic speedometer. When a star is moving away from us, this motion shifts its spectrum to slightly longer wavelength. When a star is moving closer to us, this motion shifts its spectrum to slightly shorter wavelength. This is caused by the same Doppler effect that happens with sound waves when you hear the changing pitch of a horn being blown by a passing vehicle. But the optical Doppler shift of a star moving away at 70,000 mph is only 0.01%, far too slight to be noticed by 3-coned creatures like us. We would need 10,000 distinct types of cones. Fortunately, we can build instruments that have this kind of capability.