Stanford Photo Scientists are Teaching the Camera New Tricks

Uploaded by StanfordUniversity on 26.08.2009

Somehow the name Franken Camera seems to have stuck. The idea is in fact to build a
camera that we can program right down to the microsecond level. And where we can
swap out the lenses, we can swap out the sensor, we can put other things inside the
camera that we want to play with, and we can program it easily. Existing cameras
cannot be programmed. Single lens reflex camera that you buy from Cannon or Nikon
is a closed proprietary platform. >> It looks like very custom construction it
really actually is a bunch of off the shelf parts >> In general confrontation
photography should let you fix more things that you cannot currently fix by combining
the multiple pictures that it took in different way or by fiddling with the
optic so that more is being recorded than a normal camera records. >> It’s bit too
dark on the left side if you open up the aperture that looks fine but now this is
too bright I can take those two shots and throw them into Photoshop and get this
nice result where both of those are well exposed. I want my camera to be able to do
that automatically right after I take the shot. But the main large board here is a
development board for the CPU we're using. >> But sometimes you'd like to be able to
adjust the focus after you take the picture rather than having a picture in
which, in which just one thing is in focus or in which everything is in focus. So if
you place a micro-lens array into the camera we've shown that you can re,
refocus it, change the focus after you take the picture. >> You could have also,
this high dynamic range mode, in which you are changing, setting the camera to,
[inaudible], to go to three different exposures. >> The whole field of
computational photography is, and will change, the way we do photography. The
question is, what do people work on next? So for the last ten years, there's been a
megapixel war, as everyone knows. That's finally winding down. The Asian camera
manufacturers can no longer compete based on the number of megapixels, so they are
beginning to compete on one other feature. We'd like them to offer lots and lots of
features that the research community is currently working on. And to offer those
all at once, not merely when they want to compete with one another. >> The most
satisfying result would be something completely unexpected. We would like to.
To make these programmable enough that a student in a course or another researcher
in a university can pick one of these up and create something that they couldn't
have been able to create on other existing camera platforms.