The CCD and the entire setup has many other sources of error which have a multiplicative effect, meaning a certain pixel will always be X times less sensitive than average. This can be from a variety of reasons, but most commonly caused from dust somewhere near the focus, and variations of the sensitivity of each pixel. You can see both of those effects in the image above. Each of those rings are out of focus dust particles in the system. You can also see that the right side CCD chip looks less sensitive (darker) than the rest of it. If our system were perfect, we would see a pure grey (flat) field, since this was an image of the uniform sky at sunset. The pattern we see when we should see nothing gives us the way to correct all of our real images, by dividing out this effect.

Since each filter can introduce different effects because the pixel sensativity is a function of wavelength, a flat field must be created for each filter (RGB), and clear. The way to characterize this is to take a picture of an evenly illuminated surface, like the sky at sunset, or a nice, evenly illuminated white screen.