On Low Light

Right now camera tech is moving fast and it’s fun to work with, whether it’s your hobby or avocation or profession. The big deal, of course is the arrival of the first of the long-awaited DMD (Decisive Moment Digital) cameras. The desire that drives it all, mostly, is for cameras that do more with less; specifically, less light. We’re getting them, which is good, but I have one minor regret.

The DMD notion linked above was introduced years ago by Mike Johnston, who normally writes at The Online Photographer, my personal favorite blog in this space. It isn’t hard to understand: a camera that’s small and takes pictures quickly, and has enough pixels so that people who print can make big ones, and is sensitive enough that you don’t normally need a flash.

Serious photographers or camera-tech geeks can stop here, the rest will be old hat. But I know that I follow this tech more closely than a lot of the readers here, so I suspect this will be useful for some people. And while we’re doing program notes: All the pictures in this piece were shot in natural light at f/2.0 or wider, mostly with the Sigma lens mentioned below.

On Flashes · Can’t stand ’em. Every photographer in the world hates the flat garish look you get from a flash stuck on top of your camera when you point it at someone from not too far away. There are two solutions to this, one of which is don’t use one; use the natural light that’s there. Alternatively, of course, you can become a lighting maven, and this is a vigorous and amusing sub-hobby of photography; for a taste of it, read Strobist. You can master all sorts of difficult and delicate new skills and hey, spend thousands more dollars on the gear.

Well, except for I’d rather not. When forced to flash I’ll hold a little piece of white paper up to bounce it off the ceiling, and occasionally I’ll get something that’s not awful, but you sure don’t get that decisive-moment feeling.

Shots in the Dark · So it’s really a Very Good Thing when a camera can take the picture where you are right now with the light you have right now. The first few waves of digital cameras were typically really lousy at this, because:

• The earliest didn’t have enough megapixels, so the manufacturers got into a more-is-better race, packing as many as possible onto the sensor at the heart of your camera. Except for, when you crowd ’em in like that, they become cramped, wimpy, and unhappy; fewer photons fall on each and your sensitivity suffers.

  This led to madness like Fuji, back in 2006, throwing away its primary competitive advantage so it could claim a bigger megapixel number.

• An easy way to uncrowd the pixels and capture more light is to use a bigger sensor. It turns out that fitting one of those into a small camera body is a really hard engineering problem. Not just electronics either; positioning a lens on a small body so that it can illuminate a larger sensor is very tricky.

  Wikipedia has a handy graphical key to sensor sizes. Most pocket point-and-shoot cameras these days use 1/1.7" sensors. Which is very small.

• It’s easier to capture more light with a fixed (“prime”) wide-angle lens, but consumers like the mechanical zoom lenses on point-and-shoot cameras, and for good reason: you can shoot things that are further away, and you can compose while standing still. But it’s very difficult to design a zoom that fits into a pocket camera, and works well, and lets in lots of light.

• Finally, if you need more light, you can hold the shutter open longer and just wait for it to add up; assuming that what you’re shooting is holding still. Except for, at anything much less than a sixtieth of a second, without a tripod your natural shakiness will lead to blur.

Attacking On All Fronts · The camera makers, sometime a couple of years go, seemed to notice this whole DMD concept, and have been starting to ship products which attack all of these low-light problems at once.

First of all, the insane pursuit of pixel counts is over. The most obvious example of that is the recently-announced Canon G11, which has 10MP, down from 14.7 in its predecessor the G10.

Second, bigger sensors are being shoe-horned into small cameras. Leica is making heroic strides here, with a full-frame sensor in the little M9 rangefinder and an APS-C in the pocketable X1; with correspondingly heroic prices. But probably the big news is the Micro Four Thirds system, a standard created by two competitors, Olympic and Panasonic, which allows a pretty big sensor to fit into a pretty small body.

Third, lenses are getting brighter. We’re seeing fast prime wide angles, for example Panasonic’s nifty Lumix G 20mm F1.7 ASPH for the Micro Four Thirds cameras. But even the built-in zoom on the pocket-sized Canon S90 is advertised as going to f/2.0; now, that’s an engineering triumph.

And finally, built-in shake reduction is becoming ubiquitous, so you can use slower shutter speeds. This can be done either in the lens or the body. It seems a no-brainer that in-body is better, but the Canon/Nikon duopoly have both gone for in-lens. Hmph.

All this is working. Here are a bunch of links into DPReview; high-ISO shots from the S90, and comparo pages for the Canon G11, Panasonic GF1, and Leica X1.

This essay has so far focused on small cameras; but low-light performance is being pushed even harder in the big-dog space. Of the Nikon D3 and D700 series, and the recent Canon 5Ds, it’s been said that they can see in the dark probably better than you can.

But something that you need a backpack to carry and two hands to hold just isn’t a Decisive Moment camera.

Micro Four Thirds · That Canon S90 pushes the small-sensor envelope further than I thought was possible, so apparently there’s life in that format; but there’s no doubt that the photo tribe is mostly palpitating these days over the M4/3 cameras. They’re petite, they’re slick, they come with lovely fast primes and nifty little zooms, and they have pretty good low-light performance.

There are only a handful of them, so far. In particular, the Olympus E-P1 (and closely-related -P2) and the Panasonic GF1 have been eliciting fanboy swooning from normally grizzled/cynical camera writers. And then there’s GF1 Field Test by Craig Mod; possibly the single most compelling rave camera review in the history of the universe.

Sure makes me want one; but I’m not exactly shopping at the moment, and it is after all the GF1; there’ll be a GF2 along pretty soon now...

On Fast Lenses · One of the basic low-light techniques is using a “fast lens”, which is to say one that opens up wide; anything f/2.0 or below qualifies. Now, there are downsides; these lenses tend to be big and heavy (you need a lot of glass to gather all those photons), and, when wide open, have a brutally tiny depth-of-field.

Gordon Lewis wrote on the issues around this recently in The Risks and Rewards of Fast Lenses and The Risks and Rewards of Fast Lenses: Part II; and I have my own story to tell. In October I went to Oracle Open World, shooting mostly inside, in the Moscone center. Since then, the weather here in Vancouver has been almost uniformly terrible, and my one escape was to RubyConf, where I was doing more indoor shooting.

Which is to say, I’ve been a low-light specialist, willy-nilly, for the last few months. And since my Pentax K20D has average (at best) low-light performance, I’ve spent that time photographing primarily through the Sigma 30mm f/1.4 prime. It’s acceptably — not brilliantly — sharp, weighs more than the camera body, and has laughable depth of field. It was particularly brutal at Oracle Open World; I lost tons of good pictures of interesting people because I wasn’t careful enough with the focus or exposure.

But suddenly I’ve noticed that when I go back to the lighter, sharper, smoother, Pentax lenses, I have far fewer blown bad-focus or bad-exposure shots than I used to. Which is to say, I think, that this has been a very valuable exercise. But I suppose in a few years every little camera will be able to see in the dark as well as I can, so there’ll be no need for anyone ever to repeat it.

---

---

Contributions

---