Between all the new digital cameras pooped out before the upcoming PMA show and the crazy cameras buried inside cellphones at MWC, it's a good time to go over why more megapixels isn't necessarily better.
So, the nutshell explanation of how a digital camera works is that light lands on a sensor, which converts the light into electrical charges. Depending on the kind of camera you're using, how the light reaches the sensor may seem different—honkin' digital SLRs house a complicated pentaprism and mirror system that swings out of the way, while the inside of a compact point-and-shoot is mechanically far simpler. At the heart, though, the sensor fundamentals stay the same.
The sensor is where most of the megapixel machismo comes from. When you squeeze the shutter button, the sensor (like film in old-school cameras) is exposed to light for however long you have the exposure time set for. The most common metaphor to talk about how a sensor works is that it's like an array of buckets (the pixels) that collect light, and the amount collected is turned into an electrical charge, which is converted into data. We talked a bit about the differences between the two major types of sensors, CCD and APS (CMOS) earlier.
Generally, the more pixels packed onto a sensor, the higher the resolution of the images it can produce. (Image resolution is somewhat confusingly also measured in pixels, but the term pixels doesn't always refer to the exact same thing.) A megapixel is 1 million pixels, so a 12-megapixel photo has a resolution of about 12 million pixels. Sounds like a lot, till you consider gigapixel photos, which have over a billion pixels in them. By comparison, a 30-inch monitor with a 2560x1600 display resolution amounts to a measly four megapixels, and even the best high-definition video currently is around two megapixels, no matter how large the TV.
The most recent crop of $250ish point-and-shoot digital cameras from Canon and Nikon seem to establish 10-12 megapixels as the new norm for everyday pocket cams, and hell, Sony Ericsson crammed a 12-megapixel into their tiny Idou cellphone—the same as Canon's entry-level XSi DSLR and Nikon's $3000 D700 pro DSLR. Uh, what the hell?
Obviously, there's a world of difference between the image quality you're going to get out each of those. Most of it comes down to the size of the sensor and the pixels. You can fit a much bigger sensor inside of a DSLR than you can inside of a cellphone, which not only means you can fit more pixels on the sensor, you can fit much bigger ones—imagine bigger buckets to catch the light. Sure enough, the sensors inside of DSLRs are huge compared to the ones in compacts as DPReview's detailed size chart shows. They also explain how to the read the sizes—which actually refer to the size of the tube around the sensor, not the sensor itself. Sensor sizes are referenced against 35mm film as a standard—cameras with sensors equivalent in size to 35mm film are called full-frame, though right now that's limited to pricey semi-pro level DSLRs.
To get really high-resolution smaller cameras and phones, manufacturers pack as many teeny pixels as tightly as they can onto tiny sensors. The pixels in standard point-and-shoots aren't the same kind of high-quality pixels found on DSLRs—and generally speaking, bargain bin cameras will offer lower quality pixels than higher-end shooters of the same class—which results crappier color accuracy and usually lower dynamic range too.
The other problem is noise. When you pack in pixels like delicious cows headed for slaughter, you create a lot of heat, which is one of the ways noise is generated—the rainbow colored random grain you see sometimes on digital photos. Noise gets worse as you crank the ISO, amplifying the sensor's sensitivity to light. In newer point and shoots, it's really noticeable around a sensitivity of ISO 800, though the D700 and 5D Mark II DSLRs can be jacked up to 3200 ISO and produce acceptable images (we've used some on Giz).
So, on a given sensor size, a lower megapixel count with bigger pixels will produce cleaner images—hence the D3 only rocking 12 megapixels. Most cameras mitigate noise with fancy noise reduction algorithms that are getting better all the time—Canon's 5D Mark II manages to balance delivering 21 megapixels with images about as clean as the D700's at higher ISOs—but for the most part, we're happier to see bigger sensors and fewer pixels.
One disadvantage of the bigger "buckets" in DSLRs is that you do need more light to fill them up, meaning you might need longer exposure times.
And when it comes to print quality—that old argument for extra megapixels—for most of the shooting the average person does, 6 megapixels is just fine, as David Pogue shows (and Ken Rockwell has more on), since you can make ginormous prints from it, and a clear, noise-free 8x10 looks better than a crappy one when its framed and hung on the wall. As Douglas Sterling told us via email, pros crave the extra detail of ginormous megapixel images, but when it comes down to buying cameras for regular people, just keep in mind that more megapixels isn't necessarily more better. It's how good those pixels are that matters.
Something you still wanna know? Send any questions about cameras, obscuras, or Waffle House to firstname.lastname@example.org, with "Giz Explains" in the subject line.