One of my favourite blogospherical institutions is David Thompson’s Friday Ephemera. No matter what else may be happening in the world, there, every Friday, they are. The world’s financial system may be going to hell. My life may be a perpetual disappointment, doomed in not very may years to end, probably in pain. But meanwhile, never fear, every Friday, a couple of clicks will get me to things like … a horse in a car … spiral staircases … whisky barrel flooring … the credit crunch in the form of aerial photos of Florida … a sex toys chess set … cool bookshops … a cat with bionic legs … a high rise tennis court … secure parking … an oddly shaped football pitch (that was on a Sunday but look at it anyway) … a fish with hands … bookshelf porn … Japanese travel posters … or a scary trick like this (not for those with heart problems).
Ninety five bloggers out of a hundred with a taste for such trivia would give each of these oddities a posting to itself, and add a paragraph or two of superfluous waffle (although that’s what I usually do, so maybe I am projecting there). But David Thompson is merely sprinkling a little weekly seasoning upon what is basically a very serious blog. His more typical meat and two veg posting is something like a fisking of some piece of leftist nonsense, or maybe several such pieces.
Last Friday, I had the honour of providing not one but two of David’s chosen ephemera. One was a cat seeing off some alligators, and the other was a video taken with a mobile phone from the inside of an airplane of its propeller, in motion.
I promised David Thompson that I would ask Samizdata’s notably educated commentariat to explain the strange effect with that propeller, and this is me doing that.
Can anyone say what is going on at the other end of that last link, in a way that makes it seem less than totally bizarre?
That’s an effect of the sensor of the camera. Likely a cheap video camera or a cheap smartphone camera with a rolling shutter. The entire scene isn’t captured in a single moment, rather with a rolling shutter the sensor scans across the scene progressively from top to bottom very fast, just not fast enough for a fast moving scene or a fast moving subject like a rotor.
My guess would be an artifact caused by how the camera’s firmware reads the camera’s sensor.
In a normal stroboscopic effect with as regular movie camera each frame is a valid snapshot, the problem is the sequence. The classic example being the wheels on the stagecoach appearing to be going backwards.
In a digital camera, a frame is not a snapshot, as the sensor is being read at the same time as the subject is moving. If you imagine a ping pong ball falling through a frame at the same rate the frame is being scanned top to bottom, the ball will appear as a pole. Scan in the opposite direction and the ball will be flattened, like a discus.
It not a very satisfactory answer I know, as the prop appears optically distorted, as if viewed with a corrugated mirror, rather than digitally, which I would expect to have had a chopping up effect, but it would have been interesting to see what effect rotating the camera would have had.
Brian,
I know a bit about cameras, optics and aircraft but I’m utterly scoobied. All I con suggest (in addition to Roue and Curmudgeon’s explanations) is that it is picking-up both advancing and retreating blades.
Seems to me the first two commenters have the right way of it.
Yes, thanks in particular for those first two comments. I still don’t completely get it, but I now get what it is that I am trying to get. That the scanning somehow collides creatively with the motion of the blades makes sense, in a way that regular photography said to me was impossible.
Blog and learn.
For a little further explanation, and how an iPhone camera captures a spinning propeller, here’s a link from one of the related videos:
http://scalarmotion.wordpress.com/2009/03/15/propeller-image-aliasing/
Cheap and nasty cellphone cameras could be made to produce some interesting effects…
Nick Lane’s link says it better. I had discounted the possibility the prop was going at an exact multiple of the scan rate as being too unlikely, but that does seem to be the culprit.