While most folks get real excited over a cam like the
Seitz 6x17 Digital that shoots at 160 megapixels,
Rice University researchers have decided that less is, in fact, more. Scientists at the esteemed academic ivory tower in Houston, Texas have determined a way to build a single pixel camera that they claim will be cheaper way to take pictures in the future. Using one photodiode and one digital micromirror device (DMD) -- which is used primarily in digital TVs and projectors to convert digital information to light (and vice versa) via its thousands of tiny mirrors -- light is "shined onto the DMD and bounced from there though a second lens that focuses the light reflected by the DMD onto a single photodiode." Then, the DMD's mirrors shuffle at random for each new light sample, creating a new pixel value. The pair of lenses and the DMD thus compress data from a bigger image (left) into a smaller approximation (right). That said, don't expect this technology to make your consumer digicam any cheaper real soon, as the prototype requires five minutes for the engineers to take a picture using this technique, and even then, they can only shoot still objects.
posted Oct 3rd 2006 3:40AM
DMDs are better known in their TI incarnation as Digital Light Processing, e.g. DLP projectors.
The idea of using a single photosensitive element is nothing new - drum scanners have been in use since the 1940s and use a single photomultiplier tube, with a rotating transparent scanning drum to build up the image one line at a time.
The difference here is that if we want to acquire an image of 1 million pixels, we do not have to take 1 million readings from the sensor. Rather, about 20-40% of that (and the percentage goes down as the image size increases...)
They really could have used a more appealing example subject.
So the guys in the lab kept a Baboon still of 5 minutes! I'd say that was more impressive than the camera...
Extreme sexual imagery, anyone?
lol
Sounds like a stale idea
I don't get it. Why would some tech guys have a baboon in their lab?
I'm assuming that the eventual goal is to combine a bunch of these, compound-eye style, to composite a big, high-res picture out of a bunch of little, low-res ones. In other words, rather than increasing the pixel density of camera sensors, increase the amount of picture information you can get from each pixel. Which would be a neat approach, I think, though time will tell whether it's worthwhile. (I think pixel density, amazingly, has a fair way to go yet, and the really high-end digicams are already beating film for sheer resolution, though not always necessarily for colour or exposure lattitude.)
Poor dead baboon. =*-(
This'll never get out of the lab. Can you really believe that the cost of "thousands of tiny mirrors" will somehow come out to be less than thousands of pixels when CCD costs continue to drop and densities continue to increase. This may make its way into strange scientific settings (like using one of these with a 10 megapixel CCD to create billion megapixel recordings for space telescopes or something), but this seems hopelessly complex for any type of consumer oriented device.
It's a great concept but they have taken the inexpensive part (electronics) and replaced them extremely expensive hardware (mirrors & lenses). Jar just one of the lenses a little and the blurry factor will go off the charts.
w00t, yay Rice
Hmm, Am I the only one who seems to remember a certain patent from apple that would have a very good use for this? That LCD screen with the thousands/millions of image sensors could possibly become real. Now, lets see them get this out of the lab and make it a bit better so i can spy on people using my computer monitor
badass... think of the possibilities... think of all those shitty "still" nasa pics