Single-pixel camera takes on digital

It certainly has potential, but it is only a very old idea (scanning a picture) applied to current technology.

The first lunar landing used a "spinning disk with a slit" TV (a return to the first television type technology) camera to take pictures that could be relayed back to earth with the bandwidth available at the time. And, the image orthicon tube formerly used in TV cameras also involved scanning methodology. Some current projectors use a matrix of controllable micro mirrors to create an image by scanning across a screen, not to mention the scanning in a conventional CRT (also known as a kinescope)

Their assertion is that we throw away 80-90% of our 10 million pixel camera's pixels anyway. They set 10-20% output as a normal benchmark up front. Here's what I calculated: A monopixel camera, if the shutter duration is even a quick 1/500th of a second, in order to do the work of a ten megapixel camera, would require 20,000 mirrored exposures in a row, a total of 40 seconds per image. Going to a 1/60th of a second exposure time, exposure times would be roughly 3200 seconds, four minutes. Apply 80-90% reductions to the above, and you have substantially reduced, but unimpressive, results. The monopixel camera is actually thousands of time lapse photographs recombined as if they were one moment in time, which they are not. This time differential includes color temperture and hue variations challenges, not to mention having flash units that can trigger up to 20,000 times for each photograph. A high resolution image contains more delayed action pixels, and a low resolution image contains less delayed action frames. Overall, interesting but flawed assumptions on the face of it. I suspect/hope they're just revealing what they want to reveal. In particular, let's see that one pixel which can pulse 20,000 times more than it did before, and how often does it need replacing? Is there a complimentary block of dry ice with each camera IOW???- Ed