Why Megapixels Don’t Matter…
…Or do they? Camera manufacturers today — especially consumer camera manufacturers — have created sensors that keep growing the megapixel count on each chip year after year, meeting consumers’ general expectations for a higher number of ‘MP’ with every subsequent refresh. But what do these numbers really mean? Do you need all of those pixels? These are some of the many questions discussed in the full post article.
Generally speaking, any application of gear knowledge in the world of photography can be applied to two distinct groups: consumer and professional applications. If we compare the growth of megapixels between consumer and mainstream professional camera models in the last couple years, we’d see a relatively large growth in the consumer point-and-shoot cameras as opposed to professional, full-body DSLRs. One can get a 20-megapixel compact camera today for $200, yet Nikon’s most recent flagship camera only has 16MP and costs $6000. For the price difference, shouldn’t it have at least 100MP? So what’s the difference? Hint: it’s not because it’s missing a pop-up flash.
In order to begin to understand what makes a camera ‘good’ in this context, we need to understand what a pixel is, and why megapixels do matter. Essentially, a pixel is a single measurement of light in three channels (a red, green, and blue channel). The pixel holds the information of the intensity level (on a scale of 0-255, for 256 colors per channel) of each channel at a specific moment in time (at the time a photo is taken) at that pixel’s specific location. If we were to have one giant pixel, it would be a giant square of any number of millions or billions of colors. With millions of pixels (megapixels) packed tightly together, we start to form an image (think pointillism).
With more pixels, we can do one of three things. First, we can crop images. The more pixels we have to work with, the more we can cut out certain portions of an image and still have enough pixels to produce an image large enough to view reasonably. Similarly, we can also create enlargements from images with greater amounts of pixels. The more pixels, the larger we can show (on a screen) or print (on paper) an image. Finally, editing digital photographs degrades the image file with every step. However, if not edited to excess, certain changes won’t visibly degrade image quality. Having a greater number of pixels in an image to begin with can allow for more extreme editing in post without necessarily seeing the effects until enlarging the image further. So, megapixels do matter, but only to an extent.
There’s only so much that one would ever want to (or need to) enlarge an image. If you never enlarge anything past 20×20, then at some point, more pixels only take up more space on the memory card. A few extra are nice to have so we can make edits and crop in a bit without any visible degradation, but eventually pixels just waste space — and we’re at the cusp of that point in time. For an extremely high-quality, 8×10 print, you’d only need an image that’s 2400×3000 pixels (or just over 7MP). At 21MP, you can take a photo of three faces and create a sharp 8×10 of just one (cropping in).
But if the average user really only ever uses around 10MP, then why are we building more pixels into the same cameras year after year? Aside from special applications (or the landscape or fashion photography industries), we don’t need it. The short answer: because everyone expects it.
I’ve only recently seen an honest effort on Fuji’s behalf (with the release of the X-Pro1, but more on that later) to introduce a new aspect and teach consumers by example of other ways to increase image quality.
First things first: how else do we create sharper, cleaner images? How do we apply a ‘quality, not quantity’ mindset to imaging technology? One way is by having larger pixels. Similar to the ability of your eye to dilate and increase in size to gather more light in a dark room, the larger the pixel, the greater the ability to capture light. Therefore, larger pixels lead to better low-light performance (or better ISO performance).
One way to do this would be to decrease the number of pixels on a single sensor. However, this is counterintuitive, as the industry has just spent the last fifteen years working toward an affordable high-megapixel count so we can have great enlargements.
The other way to make larger pixels is to increase the size of the sensor. Of course, manufacturers are slightly limited, as compact cameras can only hold so large of a chip. But as other electronic components shrink (and as chips become cheaper), manufacturers have more recently been able to introduce 3/4-inch and APS-C sensors into compact cameras that can fit in a pocket. Still, these aren’t quite the size of a full-frame (the common 35mm film size) sensor; but they are fast-approaching.
Another way to practice the ‘quality, not quantity’ ethic is to introduce new sensor technologies. As touched on earlier, the Fuji X-Pro1 features a new sensor technology that helps overall image quality tremendously.
Aside from the X-Pro1, every digital camera has something called an AA filter (anti-aliasing filter). This isn’t exactly precise, but for the ease of understanding, the AA filter keeps the sensor from getting tricked by straight lines and patterns (i.e. brick walls), an effect known as moire, but does so by slightly blurring the image. Fuji found a way around that with a new sensor design, allowing for the first camera in its class without an AA filter, which vastly improves image quality. For the first time, nothing but air comes between a lens and the sensor. The new Fuji camera doesn’t sport a full-frame sensor, but its APS-C-size sensor is supposed to out-perform some of its larger, full-frame cousins. New technologies such as Fuji’s will help image quality in the future in a new way entirely. Companies like Leica leave the AA filter out completely, but let users deal with moire in the few images that it does come up in. They have yet to introduce a similar technology to Fuji’s.
Fovean, a company bought by Sigma a few years back, also has an interesting technology. While Fovean-sensor cameras still use an AA filter, they make use of a different technology to eliminate moire in cameras such as the Sigma Merrill cameras we briefly discussed a while back.
We’ve all seen images taken with point-and-shoot cameras and wrinkled our foreheads as everyone points at Sarah’s dress and says, “That’s not really the red it was…you have to see it in person. It was fabulous!” This is because point-and-shoot cameras lack the color reproduction properties of professional-grade cameras — another reason people pay over $50,000 for a medium format digital camera in the fashion industry: color is nuance, and nuance is everything. Sony has a new, much-touted back-lit sensor that’s supposed to make colors really pop. Are they more accurate? I’m not entirely sure, as I haven’t tested it. But it likely improves upon color accuracy in consumer-grade cameras at least slightly.
Naturally, all of these new technologies rely on getting the image in focus to begin with. The relatively new camera by Lytro, also covered before on our site (There’s still time to win one here!), is another example of innovation that can improve image quality, albeit, in a slightly different way. Need to grab a quick snapshot? Point the Lytro in the direction of your heart’s desire and shoot away. With a resolution of 16 million light rays, whatever that means, the Lytro already sounds like something out of science fiction. But imagine this: what if you could choose your point of focus after the image was taken? No really, what if? It’s not science fiction anymore. Lytro images require a special viewer to be able to select focus and see it change on the spot, but click on any part of the image and I guarantee you’ll think something is rigged. This is a journalist’s dream come true!
When we have cameras such as the Nikon D800, just remember: those 36 megapixels are nice, for the first hundred images. But then they’re slow to work with on the camera and on the computer, they take up a ridiculous amount of space. And really — who really needs all of that? While it’s still the best deal Nikon has in that price range, I say wait on the rumored and much-anticipated Nikon D600. I think that will be the camera we’ve all been waiting for.
In all, 24 megapixels seems to be the prime number. Until hard drive space becomes so cheap and so compact that we really don’t care about another ten terabytes of data, 36 megapixels is just too much for the average user — even for the average pro, save for the few in the fashion industry. In reality, the 18-24 megapixel sensor size allows for perfect HD video downsampling (at 22 megapixels) and great cropping flexibility that even the news industry could find useful time to time, even if 12 megapixels is usually more than enough. Being able to take a portrait of three people and pull out a perfectly sharp 8×10 headshot of one of them is a great feature. But no one needs to pull out an eyelash from that distance…
Nikon has capitalized on this (or will) with its D600 and D400 FX and DX cameras, if and when they come out. Canon has gone the video route, but has still given users a great still camera with the 5D Mark III. And until we see a fusion of both for under $2000, the industry won’t change much. Thirty-six megapixels at 12fps might change something. ‘Til then, we’ll stick with what’s necessary and cut the bulk.
It’s new technologies like these that will see through the next transformations in imaging technology. The digital camera was one innovation. But while nearly all modern cameras build upon the digital format, it’s all but the last.
***NOTE: I apologize to those who viewed the version of this article naming Olympus as the owner of Foveon and saying the OM-D EM-5 camera featured this sensor. This is incorrect. Somehow, my lines got crossed, but it’s all straightened out now. Thanks for your understanding. I hope you enjoy the correct version.