The Signal vs. Noise blog had an interesting post about using flipbook-style image sets to visualize things like dance moves and movie clips in a scannable way.

This inspired me to poke around on YouTube. To my surprise, I found that they actually provide three stills from each movie, with predictable URLs (thumbnail URLs end with /2.jpg; the other frames are /1.jpg and /3.jpg). However, YouTube only displays a single still, as the thumbnail.

So I wrote a Greasemonkey script that flips thumbnails between all three images at very short intervals, essentially animating them: YouTube thumbnail animator.

Now, this isn't really the scannable sequential imagery of the Signal vs. Noise post, but it's still cool and rather useful. Perhaps a bit CPU-greedy.