Does the panel have a really good timing system that makes the lights flash as it rotates across the appropriate point?
Yes - exactly that. (Well, not just timing - the system will have some sort of sensor that will allow it to keep track of the doped of rotation,or the position of the rotating part) There are spherical versions of this concept which are quite impressive.
For minimizing cost, it may only be a single sensor that provides one tick per revolution, with a timer that counts how far into one rev it is based on how long it took to complete the previous rev.
At Brookstone you can find a clock that operates in a similar manner - except instead of spinning, the LED bar oscillated back and forth in the air, like a hand waving a wand; a digital clock face is made to appear in thin air.
Indeed, or a single sensor that measures the RPM and adjusts the motor speed up or down toward a target.
I wonder if the ‘clock’ has been added to the film digitally. It’s quite a trick synchronizing the flashes to the frames of the film, if real. Plus the fact that the whole clock is still visible on freezeframe. These things usually require persistence of vision to work, and only show a part of the image at any given instant.
If it’s a fairly cheap camera, the sensor might not have a very fast response time, which would create an effect similar to persistence of vision. I’ve seen enough of these types of display demonstrated on video not to suspect this one of even honest trickery.
Here’s a fancier version of the same concept, displaying not only a clock, but animated graphics. The site claims they have a cartesian to polar coordinate function that allows horizontal text scrolling mixed with other graphics. Color me impressed.
Peter, while photo manipulation of an image is always possible, there’s nothing here that can’t be accomplished by some pretty sophisticated software, or at least more sophisticated than a punch card reader. While I can’t claim to have ever written anything close to this, I can get a pretty good idea of what it entails and how it could be done.
And to get the virtual effect of human persistence of vision, all you need to do is adjust the camera shutter parameters until it looks right.
I wrote code to do this 20 years ago. It’s almost trivially easy. Interestingly, this techniques is (or was) patented, but the guy who patented it never made any money off of it - we talked to him about licensing his patent.
Does it serve any purpose other than being cool? It seems that a stationary flat screen would be a lot easier to program rather than plotting the movement of a rotating stick (plus having to rotate the stick). Coolness aside, is there some advantage?
Here’s a toy that I own: The I-Top. The difference is that it freely spins on any flat surface, so it doesn’t have the non-rotating base that would tell it when it’s completed exactly one revolution. It took me a while to figure out how it knows.
A hint is that one of its functions displays a magnetic compass with the LEDs
Well, it appears he’s managed to illuminate a rather large circular area using a single line of just ~50 LED’s; that’s less then half a percent of the LED’s it would take to lay out an entire circular display of the same size. If LED’s are $1 apiece, I’d say that’s a pretty nifty advantage.
Being cool isn’t enough for you? 
I can’t think of any advantages and I can imagine a few disadvantages, so as far as I can tell coolness/style is the main point.
These are often called “propeller clocks” and they aren’t that difficult to make. I’ve seen quite a few over the years.
Here are some links if you are interested. Most of these have schematics and source code:
http://www.bobblick.com/techref/projects/propclock/propclock.html
http://www.luberth.com/analog.htm
http://www.electronixandmore.com/project/propclock/index.html
You can find many more by googling “propeller clock”.
Yep, these things are spinning at 15-60 revolutions per second on average (based on what I could find on the internet), so an exposure of equivalent time or slower will capture at least one full sweep of the clock and display the persistence effect. Anybody who’s ever had to do a still capture of a video feed or computer monitor knows has seen the same basic idea in action. Shoot at a fast shutter speed (like 1/250 second), and you’ll get huge black bars in your image where the raster hasn’t illuminated yet. Shoot at 1/15 or slower, and you’re certain to get at least one full raster sweep recorded onto your still camera, and therefore, a full image.
Just to clarify, I’m not disputing that the device is real. I’ve seen that same sort of thing in real life. Not quite so sophisticated, but similar. I just wondered if the video is a simulation. And on this I note the opinions of people who are better photographers than I am.
How are they powered? If battery, how long does it last? How do you set them for the correct time?
That would be me. And I don’t think that video is a simulation, but a real device.
The videos I’ve seen are mostly breadboarded, on a test bench, so powered by AC thru a power supply. The display electronics could run for a while on a battery, but the motor would draw much more current.
If you mean how does the power get from the fixed to the moving part of the device, at least one design uses a wiper-type mechanism, and a capacitor to span a short power gap if the wiper hits a rough spot. Another design mentioned a lithium battery on the moving circuit board.
Correct time: it’s a computer – could be set any number of ways, manually or automatically.