We are developing the social individualist meta-context for the future. From the very serious to the extremely frivolous... lets see what is on the mind of the Samizdata people.

Samizdata, derived from Samizdat /n. - a system of clandestine publication of banned literature in the USSR [Russ.,= self-publishing house]

Amazing new camera that focuses afterwards

And in more ways than one.

More from the Department of Capitalism-Ain’t-It-Just-Great?!?!, in the form of this incoming email from fellow Samizdatista Rob Fisher:

Have you seen this camera that does not need to be focused?

I have now.

In fact you focus afterwards by clicking on the picture: link.

An example picture: link. Click on the raindrops or click on the building.


A review: link.

No time to read that now, but I bet they think it’s amazing too.

It’s a bit early-adopter as apart from the gimmick the pictures aren’t actually that great. But imagine this is in a very good camera with lots of megapixels. Imagine lots of dynamic range so you don’t have to worry about focus or exposure … Imagine so many pixels that you can even zoom after the picture is taken …

And if it was combined with this equally astonishing flat lens

Something tells me that this will not be my last camera.

12 comments to Amazing new camera that focuses afterwards

  • Dave Walker

    I first saw news of the Lytro a few months ago; would love to have a play with one :-).

    The flat lens is interesting; I know other researchers spent some time working on holographic lenses (a standing-wave hologram of a glass lens is a flat lens, when you think about it), as these lenses are not only flat, but flexible and have interesting properties for mass-production (make a large sheet, chop it up). Did this research direction hit a block, I wonder?

  • One way I think the Lytro might work is by having some sort of mask or lens array in front of a normal sensor. If the Lytro used a sensor like the one in the Nokia 808 Pureview (a 41 megapixel sensor which makes for a very interesting camera in an otherwise not great phone), that might increase its performance.

    I do rather like the idea of adjusting focus, exposure and composition after the image is captured. You can already do a bit of the last two by taking RAW pictures and using a large sensor like on a full frame DSLR or medium format camera.

    Focus is another trick. Also the Lytro people are promising new features in software like 3D and being able to move the perspective around.

    That scene in Bladerunner might not be so daft after all.

  • Sceptical Antagonist

    I think the best use for this camera would be for creating those lovely massive-depth-of-field shots. Of course they’d have to create software to take the sharpest details from each layer, but there are already programs out there that are not bad at that sort of thing.

    Lens aberrations are not as beastly as they once were either, especially as you can now correct a lot of them in software such as Photoshop.

  • David Crawford

    At that Lytro site they had this blurb under the picture:

    Coming soon! The Lytro camera will be available on Target.com, BestBuy.com and Amazon.com.
    And, the camera will be available in Australia, Canada, Hong Kong and Singapore!

    Hmm, which english-speaking nation, with a population of about 60,000,000 people, isn’t mentioned?

  • Patrick

    I’m just surprised they don’t call it the 20-20

  • Andrew Duffin

    I call Snake-Oil.

    If the picture’s out of focus, the information is not in there, period. It cannot be re-created, it can only be guessed or interpolated by software.

    The software may be clever, and the resolution high, but it’s only an approxmation to the picture you didn’t quite take, when all’s said and done.

  • Dishman

    Andrew Duffin wrote:
    If the picture’s out of focus, the information is not in there, period. It cannot be re-created, it can only be guessed or interpolated by software.

    I’ve got an associate who worked on either this one, or one like it.

    The imager is much more complex than a standard imager, but it does retain the information. It effectively trades resolution for focus. IIUC, theoretical maximum is close to half resolution.

  • Richard Thomas

    Andrew, you’re not completely correct. The information has been spread out by convolution with another function (gaussian is the one often used to blur in software but it’s something not dissimilar for a lens). This can be rectified by deconvolution which typically involves Fourier Transforms. With modern processing power, this is probably coming within reach (though they may be using another technique).

    Some information *is* lost but there’s definitely some scope for sharpening. Though I’m not necessarily convinced this particular camera isn’t just a gimmick. I recall reading a while back of an algorithm that could do multiple tricks with fractal processing, in effect, creating information which isn’t there. More details will need to be forthcoming.

  • Well I guess the first commenter here who buys one can report back.

  • Sceptical Antagonist


    IIRC from seeing this tech previously, the sensor is not flat but 3D, which is how the device manages to capture the focus points at different depths. The sharpness is not interpolated but captured as is, albeit with a reduced resolution.

  • This really does work — no information is “created” out of thin air. It works by having many lenses in an array. Behind each lens is a number of sensor pixels. Focusing is a matter of deciding which pixel to use from each lens. It’s called a plenoptic camera and is a real thing. This is just the first one available to consumers.

  • Sceptical Antagonist


    Thanks for the explanation – it wasn’t how I remembered it!

    I had it in my head that there were layers of sensors with gaps in between the pixels. Must have misinterpreted a diagram or something.