On Megapixels

Suppose you’re interested in buying a camera. If you look at the ads and reviews, the first thing you see right beside every single one is its megapixel count. The camera makers want you to think that more is always better, which is wrong. But the community buzz is starting to be “more is worse”, which isn’t really right, either.

Why More Used To Be Better · I was an early adopter of digicams, buying a 640x480 Fuji in 1998. I still have what I then considered the keepers, right here on this laptop; and most of ’em look like complete dogshit. Lack of talent and application are major contributors, but lack of pixels was a real issue too. Here are two of the best-looking ones, more or less (I think) as they came out of the camera.

I got my first “serious” digicam in 2003, a Canon S50. The jump to a big five megapixels was huge, life-changing. I remember shortly after I met and photographed Dave Shea, I sent him one of the shots and he wrote back “Ooh, I have serious megapixel envy”.

So from there on in, it was pretty easy: Cameras had a number attached, and bigger was better, so all the world’s camera companies’ marketing departments got into lockstep and kowtowed before the mighty megapixel count. It’s still happening... pull some flyers out of your weekend paper and check the camera ads.

Why More Can Be Worse · Well, but once you get past six or so megapixels, the trade-offs start to get ugly and unsatisfying. Basically, almost all pocket cameras have about the same size sensor (SLRs have a bigger one which is why they take better pix), and the engineers can sure enough squeeze more pixels into that space, but there’s no free lunch; those pixels are squeezed and light-starved and undernourished and so you get noise problems and sensitivity problems and so on.

Here’s Micah Marty’s great little essay Digicam Appreciation, about how one camera company (Fuji), which had a lead in making high-sensitivity cameras that took better pictures indoors, pissed it away because they had to have that big megapixels number.

Indeed, it has come to be almost a consensus, among photographers who care, that big megapixel numbers are bad not good. It’s indicative that the Nikon D3, which right at this moment in 2008 is probably the best camera in the world (but damn it’s big), has “only” 12 megapixels.

Here’s another anecdote. Recently, Ricoh announced the GX200, an update of the GX100 that lives in my pocket. It fixes the GX100’s #1 problem: slow write speed on RAW files. But, it also increases the megapixel count from 10 to 12. The reaction over at the DPReview Ricoh forum began with:

More Megapixels :-( But faster buffer for RAW :-)

And many more joined in about the “megapixel chasing” — follow that “reaction” link above for some instructive moaning.

What About Printing? · If you step outside the world of marketing brochures and hang around the online community of twitchy photography obsessives, whenever megapixels come up the context is usually printing, and the narrative reads like this: “Well, X megapixels is fine up a print size of Y.” The highest-quality prints in the professional-photographers’ space are those required by glossy magazines, which leads to some surprising values for X.

Me, I don’t give a flying fuck about printing. I look at pictures almost exclusively on the screen of my computer and if they look good I’m happy. I use larger-than-average screens, but still, the number of pixels I can squeeze into an ongoing page is immensely smaller than the number of pixels a modern digicam captures. So why should I care in the slightest about printing, or, consequently, about megapixels?

At this year’s Northern Voice, I ran into Matt Mullenweg and his brand new D3. Obviously we started talking about cameras, and when told me that he shot JPEG not RAW, and not even the highest quality JPEG “Because it’s for the Web”, I was shocked. On the one hand, this is egregious mis-use of a great camera, but on the other, Matt has a point.

More Is Better · Well, these days, I shoot with the 10-megapixel Ricoh out-of-the-pocket, and the 14.6-megapixel Pentax K20D when I really care. Both of those cameras arguably squeeze in more megapixels than is reasonable for their sensor sizes, but both have managed the engineering to get pretty decent image quality. And you know, I’ve started to adopt the (heretical) opinion that more really is better.

Here’s why: composition isn’t something you do in real-time with the camera, any more. It’s what you do with Lightroom (or equivalent) later on, and the important thing is that the camera capture everything important with good enough fidelity that, after you cut out what’s not important, what you have still looks good. As evidence, I offer some Lightroom screen grabs of pretty extreme cropping in action on my recent rose pix.

I’ll freely admit that digital cameras have made me sloppy. When I’ve been working through my inventory of film scans, I’ve noticed that with a film camera I put a lot more work into composition and leveling and so on; most of those scans I’ve published are (modulo dust and noise removal) about what came out of the camera.

Who cares? These days when I shoot, the over-riding priority is not to miss anything; I know I can fix the composition later. But only if I have enough high-quality pixels to work with.

---

---

Contributions

---