Video camera is on a tripod, zooms in, scans the scene... Software then 'adds' all the footage together and produces an ultra high quality image. A bit like a long exposure this technique would not work for anything but static shots (Even trees moving would confuse it) But, then again you could ignore people + cars - producing people-free city scapes.

As an extention - using a video camera pointing at a Flat mirror (which then moves around) you could use one camera to take high-res images of the scene from 4 angles (+/-). If you're subtle with the mirrors, and had sufficient camera quality these could be taken symaltaniously. This would work well for a time-lapse where you only have to buy one camera but get 4 films out of it.

[edit 1] responce to link "sounds a bit like this'"- my idea is the opposite! but that's interesting. Summary: trading a camera's high-resolution for low quality high-speed video. I'd note that high-speed video requires a lot of light, and in that case, you've got a number of options. One is to just use 40 very cheap web-cams and stich the footage together. (cant find link)

[edti2] sorry, its not fixed. The tripod is (ideally) computer controlled - but even panning by hand should work. It's better if the thing is ON a tripod - so the perspective does not shift TOO much, and the more sophisticated the tripod the easier the software is to write. Not fixed though, sorry.

[edit 2] also, it might be able to get higher resolution than even the zoom of the lens + the camera resolution by deducing finer details through blur + moving the camera a little. Using clever software which i'm sure exists... Else it's worth half-baking.
-- nicholaswhitworth, Apr 02 2010

more data =/= higher resolution, it simply means a large blurry file. Yes image processing can composite multiple frames in a video, to build a higher detail image by inference, but in a steady shot it would be worthless.
-- WcW, Apr 02 2010


Hmm, I do believe this might work. [+]

[WcW]//image processing can composite multiple frames in a video, to build a higher detail image by inference//

Is that not precisely what [nicholaswhitworth] proposes?

//but in a steady shot it would be worthless//

It's not a steady shot though, is it? As the lens zooms in, each pixel of the imaging device will be scanning along a line from its starting point, towards the centre of the image.
-- BunsenHoneydew, Apr 06 2010



random, halfbakery