[Philip Lelyveld comment: this story brings more technical perspective to this important tech announcement.]
[by Matt]
Lytro Inc. (Mountain View CA) has announce a point-and-shoot light field camera targeting the consumer market. No details, such as price, availability, resolution, camera size, etc. were available for the camera, however. The press release from Lytro said vaguely the camera would be available “Later this year.”
“This is the next big evolution of the camera,” said CEO and Founder Dr. Ren Ng. “The move from film to digital was extraordinary and opened up picture taking to a much larger audience. Lytro is introducing Camera 3.0, a breakthrough that lets you nail your shot every time and never miss a moment. Now you can snap once and focus later to get the perfect picture.”
Light field science was the subject of Dr. Ng’s 2006 Ph.D. dissertation in computer science at Stanford, which was awarded the internationally-recognized ACM Dissertation Award in 2007. Dr. Ng’s research focused on miniaturizing a roomful of a hundred cameras plugged in to a supercomputer in a lab. In 2011, the Lytro team will complete the job of taking light fields out of the lab and making them available in the form of a consumer light field camera.
Computational photography using light field reconstruction has been a research topic for a number of years. One of the problems with computational photography is the very large amount of data associated with a high-resolution image. To get the image quality a consumer associates with a normal 1MByte snapshot from an 8Mpixel camera, it may be necessary to store as much as 100Mbytes of data and use an imager with 800Mpixels. Obviously, this would not be practical in a point-and-shoot camera, so Insight Media is looking forward to seeing how Lytro solves this problem. Even at a professional level, 800Mpixel sensors aren’t really practical which is why researchers into computational photography have used a room full of a hundred individual cameras in the past.
Typically, both the sensor and the “lens” in computational photography are large in area but can, at least theoretically, be very thin. Since no specifications on the camera are available, it is not clear if the proposed point-and-shoot camera is also a pocket camera. Size isn’t necessarily a catastrophic barrier-people have accepted the 10″ size of the iPad, for example, to get features and a display not available in a 4″ smartphone. A 4″ – 10″ diagonal would be a reasonable size for a computational photography camera that promises to generate 3D images, as Lytro does.
Another problem with computational photography is that it doesn’t produce a viewable image until after the “computational” part. Presumably, any handheld camera from Lytro would include the basic software needed to produce an image visible on the camera display. Typically, computational photography involves post processing and image editing. Again, from a consumer point of view, is this what they want? Taking a photo but being unable to view it in its full glory except after a half hour or so of optimizing it on your computer is not really what point-and-shoot photography is all about.
Lytro has an on-line picture gallery of Adobe Flash photos that can be manipulated over the web to simulate what a consumer can do with his own computational photography images. While it is not stated, presumably these photos were generated with the Lytro camera, either a laboratory model or a prototype of the consumer version. (Note: anyone who can use a Nixie tube display as the subject of a photo demonstration is a geek after my own heart!)
Computational photography is normally based on multi-aperture imaging, as was discussed recently in Display Daily. An expanded version of this story with the available details on Lytro’s business plans will appear in the upcoming edition ofMobile Display Report.
See the original post here: http://displaydaily.com/2011/07/13/lytro-announces-light-field-camera/
