Speedy Cam

This article on Engadget, which credits this article from the New York Times, and this article from MIT seems to use a somewhat misleading phrase. To be fair, it seems like the part that I don't like comes directly from the MIT article:

at a rate of one trillion exposures per second

This sounds like they are saying that this camera is taking one trillion frames/exposures in a single second. However, if you read the actual source material they are really doing something quite different.

effective exposure time of each frame is two trillionths of a second and the resultant visualization depicts the movement of light at roughly half a trillion frames per second

So, rather than taking a trillion exposures in a single second, they are taking single exposures lasting two trillionths of a second. This would result in the 'trillion exposures per second' if they were doing it over and over a trillion times in a single second, but they aren't. Or at the very least, it isn't that simple.

The key difference, which is at least mentioned in the articles as well, is that the effective frame rate of the final visualization is taken from multiple runs. More complicated than that, each run is taking an effectively 1D capture. So the second dimension of the video is produced from multiple takes. The practical implication here is that you can't use this technique to capture anything that isn't exactly reproducible, because you need many identical takes of an event to produce something usable.

The technique seems fascinating, and regardless of my qualms with the reporting, it seems like a great accomplishment.

I love this take from the New York Times article:

If a bullet were tracked in the same fashion moving through the same fluid, the resulting movie would last three years.

Of course, because the bullet might not follow a perfectly reproducible path, this technique probably wouldn't really work. However, for perspective on just how slow this slow motion is, I love it.