Light Field Cameras - Focusing on the Future
Room to Grow
The Illum is a step in the right direction, but there’s plenty of room for improvement. At $1,599, it's as expensive as a decent system camera plus a solid lens, yet many aspects of the image quality are just not on par with cameras at even one-third of the MSRP. Edges are soft, colors are desaturated, and there’s a general graininess to the shots. If it matched the IQ of, say, an entry-level Micro Four Thirds camera, the cost would undoubtedly be easier for a lot of consumers to stomach.
Even though 5MP fills the display of even the highest-resolution laptop screens, this resolution is still inadequate for making large prints. This is a sticking point for some photographers, and others simply don’t see the point in paying so much for a camera that refocuses photos when their cellphones seem to offer the same feature for cheaper. Some just think that interacting with photos at all is a gimmick.
|The Illum still offers a constant F2 lens, but now with a more practical 30-250mm equivalent range.|
Many of those are complaints about the Illium miss the point. The Illium is not a conventional camera, and although it's true that a smartphone's ultra-fast multi-frame focus bracketing can provide a version of post-capture focusing, it has limitations. The confusion is understandable, because photo enthusiasts are entrenched in a medium that hasn't really changed since the advent of the digital SLR - or, arguably, long before that.
The problem is that these people are the natural audience for any new camera, so of course Lytro wants to get their attention. But the message is muddled. They're trying to frame the possibilities in new terms, like megarays, light field, and living pictures. That's because those concepts actually are different than todays megapixels, light without directional info, and regular photos. But it's like a new dialect, and regardless of the obvious theoretical appeal of refocusable pictures, it's tough to communicate with a core of enthusiasts who are used to thinking about photography in long-established terminology.
A common misconception is that light field cameras can only offer very limited resolution in order to provide lots of refocusing range. But the fact of the matter is that it's actually an adjustable tradeoff made at the hardware level that determines how much spatial resolution you give up in order to gain a certain amount of refocusing range.
'What Lytro does is that they are trading off the spatial resolution to get more angular resolution', says Kartik Venkataraman, CTO at Pelican Imaging, another company that works on computational photography and light-field array cameras. 'The more angular resolution you get, the better depth you have. So they’re at one extreme of that. Unfortunately that trades off the resolution quite a lot to be able to capture the light field and refocus the way it does. Maybe it’s not a level of tradeoff that people want to get that ability' (We spoke several weeks before the announcement of the Illum, and Venkataraman was referring to the original Lytro camera. But the new model uses a similar ratio of angular-to-spatial resolution).
|A cross section of the Pelican PiCam light-field smartphone camera module.|
Pelican makes a light field camera module called the PiCam, designed for use in smartphones (though it hasn't been used in any commercially available models yet). The reference design is a tiny 4x4 array that captures 8MP images plus depth information. Though it doesn't capture nearly as much depth as either Lytro model, it's enough for some refocusing capability, as well as extended depth-of-field photos, where everything is in focus.
The same concept could easily apply to higher-end cameras: More megapixels, but still some amount of refocusing ability. Genres that rely heavily on focus tracking could stand to benefit from a camera with this technology, as could styles that favor shooting with fast apertures.
Years from now, we’ll reach to a point where there's effectively no tradeoff between depth and flat resolution. All it requires is a sensor with sufficiently high resolution.
'What folks have been finding is that it’s not really the size of the pixel that determines the quality of the picture, as much as the quality of the silicon and the size of the sensor, and then the resolution', says Ren Ng, founder and chairman of Lytro. He cites the module used in the Samsung Galaxy S4, among other phones. The pixels are tiny, at just 1.1 microns, but the image quality is considered to be quite good.
Fitting those pixels on a larger die could result in a sensor north of 500MP. Ng says, 'Is there a point to that in 2D? No, definitely not. Lenses can’t resolve it and you can’t hold your body still enough to shoot that. But light field devices that capture [that resolution] - that’s where you get no tradeoffs compared to [conventional] digital cameras'. That’s still years away, but Ng says, 'it’s technology that sits nicely on the Moore’s Law growth curve'.
While refocusing is the most relatable, immediate feature of light field photos, the technology has a huge impact on lens design. The Illum uses a 30-250mm equivalent F2 lens. That would be exceedingly difficult to design for a conventional camera, requiring several aspherical elements to correct for aberrations. 'With a light field sensor, we can re-sort the light rays from where they went to where they should have gone' using only software, Ng says. Lenses for light field cameras, like the glass on the Illum, use fewer elements so that they can be smaller, lighter, and cheaper to manufacture, all while achieving longer zooms and faster apertures.
Coming into Focus
The Illum demonstrates one particular application of light field technology in photography: Capturing tons of depth to create interactive images. It won't appeal to all photographers, and to frame the Illum in terms of its limitations is missing the point.
Even Lytro knows this camera has a limited audience. Rosenthal estimates it’s somewhere between 5 and 15 percent of high-end camera buyers. Ng says it's meant to be one camera in an artist's bag - not the only camera - and it's certainly no replacement for a DSLR.
That begs the question: Is there enough niche interest to sustain interactive photography? Ng says, 'When digital cameras first came out, it was sort of like a 'Well, what exactly do I do with this VGA, 0.3-megapixel thing? Is it sort of pre-scanning my photos?' But it really led to a transformation'. Before that, photographers were skeptical about adopting color film. And long before that, portraitists were skeptical about trading their paintbrushes and palettes for cameras.
|The Illum has 40 megarays of angular resolution, providing output of about 5MP of spatial (or two-dimensional) resolution.|
However, it’s plausible the 'living picture' might fail to stick. Most photographers won’t have a practical use for a camera that creates interactive photos, unless they’re interested in working in what’s essentially a new medium. If it doesn't catch on with viewers, then there's little incentive for widespread adoption.
It's still laudable Lytro is pushing light field photography forward. All photographers - even those who can't see the Illum as more than a gimmick, and may never buy a Lytro product - will benefit from their advances. Simplified lens design is reason enough for other camera companies to start pursuing light field tech.
The really exciting stuff, though, will happen beyond the bounds of traditional stills photography. 'If we're successful, then over time, just about anything with that has a sensor with a lens in front of it will incorporate computational light field photography', says Rosenthal.
Ng brings up the possibility of a light-field microscope. It turns out that microscopic specimens are transparent because of their small size. 'All the light rays pass through to some varying amount of absorption, he says. 'You capture a light-field microscope shot, and you can compute a full three-dimensional volume from the one picture', It’ll be like the difference between an X-ray and a CT scan.
Video is the most obvious beneficiary of light field technology. Hollywood movies could be focused entirely in post production. And since light field cameras capture 3D data through a single lens, it'd be much easier to shoot 3D films - all of the parallax problems presented by stereoscopic cameras would disappear.
And then there’s the holy grail of imaging: holography. 'Light fields, every ray, traveling in every direction, every point in space, it turns out, is a hologram. So from one light field shot, you can compute a full hologram', Ng says.
We’re a long way off from that - we’ve yet to shoot the first light field video, mainly for a lack of processing power. We also don’t have a great way to properly display holograms yet, though researchers at MIT believe they could be in living rooms within 10 years.
There's a bright future for light field imaging. Even if early cameras like the Illum don't yet have a clear purpose, they're helping bring the possibilities into focus.
Liam McCabe is a technology journalist based in Boston. He's covered cameras and the imaging industry since 2009, and also knows a few things about the wild world of home appliances. On the weekends, you can find him wandering outdoors.