A little more information about the camera that takes a picture with all depths of field at once and lets a software program play with focus afterwards (and it’s apparently called a plenoptic camera): with a 16-megapixel sensor and 90,000 microlenses in front of it, you get a final 90,000-pixel photograph.
That kinda explains how the thing works (I guess each of 175 pixels associated to each lens records a different “vision” of the final pixel), and also why it’s not gonna be seen on the consumer or even prosumer markets for a few years. Not only is it too expensive for 90,000 pixels, it’ll also be a while before the microlenses array can be miniaturized enough to get a decent resolution.
Still, it’s a fascinating idea. And I wonder if the primary practical application shouldn’t rather be high-end video cameras: no jumping autofocus when there’s movement, you choose your subject in post-production!