Years ago a reader emailed me about plenoptic cameras, also known as light-field cameras, which allow an image to be refocused after the picture is taken. Sometimes referred to as a 4D camera, this crazy technology is now headed to a consumer camera from new manufacturer Lytro. News of this development, which utilizes technology first seen in a 2005 Stanford research paper, hit the internet last week, with Lytro now taking reservations for the device. Check out the refocusable images in action, and let me know what you think -- game-changer or gimmick?
First off is the company's pitch video:
And here are a few images from the company's full gallery; click anywhere on these images to refocus:
Here's an explanation of this form of computational photography in an article at The Economist:
Dr Ng's camera recreates the light field thanks to an array of microlenses inserted in between an ordinary camera lens and the image sensor. (Dr Ng declined to reveal the precise specifications for the commercial device, but prototypes from his academic days sported 90,000 minuscule lenses, arranged in a 300-by-300 grid.) Each microlens functions as a kind of superpixel. A typical camera works by recording where light strikes the focal plane—the area onto which rays passing through a lens are captured. In traditional cameras the focal plane was a piece of film; modern ones use arrays of digital sensors. In Lytro's case, however, light first passes through a microlens and only then hits the sensors behind it. By calculating the path between the lens and the sensor, the precise direction of a light ray can be reconstructed. This in turn means that it is possible to determine where the ray would strike if the focal plane were moved.
This technology obviously has a lot of potential for professional applications, but no word on higher-end implementations at present.
Why? Because, as the article brings up, the resolution of the camera is limited, presumably coming in far lower than current digital cameras (the resolution is limited by the number of microlenses, each of which produces one pixel of the "final" image). As a result, "the new device might just reignite the once-furious race for ever more megapixels." With the added depth element in the image, a twenty megapixel sensor will output an image with a much lower horizontal and vertical resolution.
This brings up another question: what about for video? Presumably the data throughput would be tremendous -- not to mention the processing power on the consumer's end in terms of decoding such a signal -- but imagine watching a movie and focusing on what you want right now instead of what the director wanted when it was shot. I'm not saying that would be better -- but it would certainly be different. Hook up such a technology to an eye-tracking mechanism that allows viewers to focus with their eyes instead of a mouse or touchscreen, and heads may explode.
For more on the camera and company, which is being hyped as "the first time picture-making has been changed since 1826," here's an interview with Lytro's founder, Ren Ng:
As Ng says in the video, pricing and availability are not public at present, but the cmaera should be out by the end of 2011 at a "competitive consumer camera price."
I'll ask again -- what do you think, game-changer or gimmick? What other applications could you see for light-field cameras?
Link: Lytro
[via FilmmakerIQ]
