A Picture is Worth a Thousand Pixels

Every digital image you take is black and white. Every one. That cheap disposal Kodak camera? Black and white. That fancy Nikon D3 you blew your paycheck on? Black and white. By their very design, all digital images are black and white. It is only by sophisticated (and respectable) bootstrapping that you can capture the red and orange frosting of your birthday cake or the turquoise Caribbean Sea for all the fans of your Picasa web album.

At the heart of every digital camera, from your Olympus to the Hubble Space Telescope's WFPC, is a Charge-Coupled Device (CCD). CCDs have remade modern photography, and before that they remade modern astronomy. They are in the process of remaking modern medicine. Without them we'd still be using glass plates to image galaxies and nebulae, and you'd be using film (which really can register color) in your Polaroid to capture poorly-lit, red-eyed memories of your vacation to the Bahamas.

So what is a CCD? We can all thank Albert Einstein for explaining it to us. Or rather, for explaining to us the the main mechanism that makes them possible: the photoelectric effect. Imagine shining a light on a metal plate. OK, there isn't really much to see. But on the atomic level, you've created a maelstrom of charged particles seething on the surface of the plate. For our purposes here we're going to think of light as a bunch of massless particles that carry momentum (photons). When the photons strike the metal surface, they have a habit of exciting the metal atoms' electrons into higher energy states. Some of the electrons get knocked out of the atoms entirely. Each metal is different, and each metal reacts differently to different wavelengths of light, regardless of intensity. Shine a red light as bright as the Sun on an aluminum plate and you get nothing, but a candle flicker's worth of blue light can create a tsunami of electrons. CCDs harness this photoelectric effect and put it to good use.

There are some metals (semi-metals, really, usually called semi-conductors) such as silicon and germanium that are especially easy to work with, and readily produce these excited electrons. They function at many wavelengths of light, whereas traditional metals may only work in specific bands or at no useful wavelengths at all. Also, we have lots of experience stamping electronic components onto semiconductors (thank you, Intel). A wafer of silicon is imprinted with a series of potential "wells," forming an enormous grid on the surface of the wafer. At its most fundamental, each well is what makes up a pixel. When you expose the silicon wafer (the CCD "chip") to the target of your observations, the camera optics focus the light on the chip. For every photon that impinges on the chip, one electron gets knocked out of the silicon substrate and falls into the potential well. At the exposure continues, more and more electrons fall into a particular well. Bright areas of your target (say, a star) get projected onto a small component of the chip, and those corresponding wells fill up with more electrons than neighboring wells. When you finish taking the picture, the shutter closes, the wells stop filling, and (through a complicated process) the camera counts all of the electrons in each well. The count of electrons gets assigned to a particular pixel, and combining all of the pixel counts together generates a full image.

Notice that color never came into play here. Without a filter, there is no way to know if an electron fell into a well after interacting with a red photon or a blue one. The CCD can only count the total number of electrons in a particular well, generating shades of grey.

Your digital camera at home works exactly the same way. To get the color, the manufacturer adds a filter layer on top of the CCD chip. A complicated (and proprietary) algorithm then sorts the red, green, and blue pixels into a color image. Sometimes there are three CCD chips and three filters, and the optics break the incoming light in three directions. There's a problem here though: both methods allow only one-third of the light to be seen for any particular color. That means you need longer exposure times. Also, the quality of the color depends entirely on the quality of your algorithm and the robustness of your filter. If you've got a filter permanently integrated to your chip and it goes sour, not only do you have bad color data, there's no telling how much light you're losing in the bad filter. And light is money, when your multi-million dollar mission depends on the images you send home!

It is also a tricky question to worry about "true" color in space. Most CCDs operate best in the red and infrared, and poorly in blue. A friend of mine does research with infrared lasers. Your eyes can't see in the infrared, which makes his lab look kind of boring even when the laser's up and running. It's just a strip of translucent fiber optic cable. But when I once took a picture of his laser with my Kodak, what did I see in the image? Bright light coming from the laser! But I though the Kodak camera only showed true color! True color is all relative. The Kodak camera is sensitive to the infrared light from my friend's laser, and in a controlled dark environment (to an extent similar to what you'll have in space, too) you get a whole bunch of light in the image that you can't see with your eye. The only "true" observer of true color is the human eye, and trying to force a CCD camera to copy it is a waste of time and effort. Indeed, it would be taking away from the extra information you can learn from other wavelengths. That's why every camera on a NASA spacecraft is black and white: better resolution, better color control through filters, and better science.

Joe Gangestad