The Frames I Skipped

I rebuilt a video from a fraction of its frames and got a calmer film than the one that existed. The energy was hiding in what I skipped.

I had a short video I wanted to rebuild, about ninety seconds of motion I needed to take apart and reconstruct piece by piece so I could understand exactly how it worked. The way you do this is you break the video into still frames and study them in order. A video is only stills shown fast enough to fool the eye, so if you have the frames, you have everything.

Ninety seconds at full speed is a couple thousand frames, and that is a lot to pull and stare at. So I cut a corner that felt completely reasonable. Instead of every frame, I took roughly every third one. Eight stills a second instead of twenty-five. Plenty, I figured, to see what was going on. I studied those, rebuilt the piece faithfully from what I saw, and sat back to watch my version next to the original.

Mine was wrong, and it took me a minute to see how. It was not missing anything I could point to. Every element was there, in the right place, doing the right thing. It was just calmer. Slower. The original had a snap to it, a quickness, a kind of nervous energy in the cuts. Mine was the same film on a sedative. Pleasant, and lifeless.

The energy was in the gaps

The thing I had lost was living in the frames I never looked at. The fast little moments, the snap of a transition, the half-beat where something flickered and was gone, all happened between my samples. By looking only eight times a second, I had not merely lost some detail. I had smoothed the whole thing into something more relaxed than it really was. I pulled it again at full speed, every frame this time, and the tempo came right back. The snap had been there the whole while. I just had not been looking often enough to catch it.

I think about this constantly now, far outside of video, because the trap is everywhere and it is sneaky in one exact way. When you measure something too rarely, it does not warn you. It does not come back blank or throw an error. It quietly hands you a calmer, smoother, more peaceful version of reality, and that version is convincing precisely because nothing in it looks wrong. A number you check once a day cannot see a problem that flares up and burns out before tomorrow. A nightly average will tell you the day went fine on the exact day the lunch rush was on fire for forty-five minutes.

So when a measurement tells me everything is calm, I have learned to ask a second question before I believe it. Not only what does it say, but how often does it look. Calm and barely-watched produce the same readout, and they are not the same thing. Most of the false peace I have ever bought came from measuring something too rarely to see it move.

/ar/