My camera-after-next will have this…
July 5th, 2006 by Jerry
Check out this plenoptic light-field camera. (Thanks for the link, Mark!) Wading through the techno-photography jargon reveals that this prototype can take one “photograph” that can be refocused at many different focal lengths. Lost a great shot because your camera focused on the wrong spot? No problem, just rerender it at the proper distance. Want a wide depth of field without having to use a narrow aperture? Also not a problem.
It looks like the only hardware modification is some kind of microlens array in front of the sensor. Some special software does the rest. Don’t ask me how it all works – I haven’t a clue. All I know is that Canon better license this technology and make a production model. 
If Nikon does it instead, I could live with that.
I won’t say my next camera will have this technology – I may eventually end up with a 30d or whatever Canon comes up with next. The next one after that better have it, though, or I’ll be quite miffed.
Update: After reading through the papers, it appears there is a down-side. The camera works by putting a bunch of tiny lenses in front of the sensor. This makes the sensor capture a bunch of small circular images. Each of these circles contains a tiny picture of the scene in front of the camera, and somehow each circle becomes a pixel in the refocused image, after it decides what part of each circle to use. I think each circle has a diameter of 12 actual pixels, and since their sensor is 16 megapixels, the final images are roughly 300×300 pixels. Not exactly high print quality. Scaling that resolution up to “normal” would require ridiculously high-resolution (and high-priced) sensors.
If Nikon does it instead, I could live with that.
I won’t say my next camera will have this technology – I may eventually end up with a 30d or whatever Canon comes up with next. The next one after that better have it, though, or I’ll be quite miffed.
Update: After reading through the papers, it appears there is a down-side. The camera works by putting a bunch of tiny lenses in front of the sensor. This makes the sensor capture a bunch of small circular images. Each of these circles contains a tiny picture of the scene in front of the camera, and somehow each circle becomes a pixel in the refocused image, after it decides what part of each circle to use. I think each circle has a diameter of 12 actual pixels, and since their sensor is 16 megapixels, the final images are roughly 300×300 pixels. Not exactly high print quality. Scaling that resolution up to “normal” would require ridiculously high-resolution (and high-priced) sensors.