The Megapixel Race Over? Hardly.
February 9, 2010
For example, he once ran a demonstration showing that random viewers couldn’t see much difference in a row of enormous, 16 x 24″ prints, even when the pixel counts varied wildly.
But Pogue made an odd aside last week, at the conclusion of his compact-camera buyer’s guide:
“As the ridiculous megapixel race winds down at last, …”
…a comment which left me scratching my head in confusion.
Perhaps he’s been busy—avalanched under press releases for all those new tablet e-readers. Or maybe he’s aggravated that the megapixel race didn’t stop at 7 Mp, as he hoped in 2006?
Believe me, I understand the frustration; the desire to throw up your hands, declare victory, and retreat.
But in reality, the megapixel war still rages—most obviously among point & shoot cameras. (And it’s the buyers of these mass-market models who are most likely to take advice from newspaper articles, rather than from some specialist geek website.)
Now, Pogue begins his compact-model roundup by noting some limitations inherent to all small cameras: shutter lag, grain, and blown highlights. But he hasn’t much followed his own oft-stated advice: Choose a camera based on its sensor size, not pixel count.
Seven of Pogue’s nine selections have pixels smaller than 1.54 microns. (The Nikon’s are a ludicrous 1.43 microns.) His Panasonic pick does a smidge better, at 1.56 µm.
But compare these to the 2.0 microns of the (still fairly compact) Canon S90—each of its pixels can collect about 70% more light.
His one choice I might grudgingly accept is the 10 Mp Fujifilm F70EXR. Besides having 1.77 micron pixels, this model offers a special low-light mode. Ironically, it works by pairing up pixels, turning it into a 5 megapixel camera! Hey, it’s a Pyrrhic victory, but I’ll take it.
But Pogue’s other picks simply have pixels that are too small, by any reasonable criterion.
I do admit that anyone forced to buy a compact digicam today—lets say your old one just died, and you’re leaving on a trip tomorrow—faces very limited choices.
If need be, you might hunt for a model using one of the new generation of 10 Mp back-illuminated CMOS sensors. For example, Sony’s “Exmor R” chip (versus the regular, non-R kind) works some special tricks to wring the most out of its 1.7 micron pixels.
This is actually rather worrying. Aren’t “enthusiast” photographers supposed to know better? That smaller pixels compromise other aspects of performance, like dynamic range and noise?
Stuffing 18 million pixels into the same 22.3 x 14.9mm sensor area makes each pixel 4.3 microns wide. This is the same pixel pitch that causes Micro Four Thirds cameras to struggle with noise when pushed up to ISO 800.
Consider the 12 Mp Pentax K-x, praised for its high-ISO performance. It uses 5.5 micron pixels instead. This gives each pixel 63% more light-gathering area.
Also remember that on the T2i’s sensor, each millimeter of sensor width contains 232 pixels.
But it is very rare for a real-world lens to resolve detail at that scale with reasonable contrast. If one can do so, it will only be at a single, optimum, middle f/stop. That’s not especially practical.
(Aberrations limit sharpness at wide f/stops; diffraction creates blur at smaller ones—in APS-C cameras, typically f/8 or smaller. For a more technical discussion, start here.)
I wish we could say that megapixel marketing madness had finally ended.
But I’m not seeing any evidence this is true.