I Made Everything Move, and It Read as Dead
Three videos came back with the same note — it felt like reading, not watching. I’d put motion into every single frame. That was exactly the problem.
I shipped three videos and the note that came back was that they felt like reading, not watching.
I didn’t argue with it, but I didn’t believe it either, because I had put movement into every single frame of all three. A slow camera push on every beat. A waveform scrolling the whole way through. Every word lighting up as it was spoken. There was never a moment where nothing was happening. If anything, I thought I might have overdone it.
So instead of arguing, I measured. I took a film I admire and measured how much the picture changes from one frame to the next, across its entire runtime. Then I did the same thing to mine.
The reference film: a typical frame-to-frame change of 0.12, and at its busiest moment, 18.1.
Mine: typical change of 0.95, and at its busiest, 1.6.
Read those two lines twice
It took me a while to actually sit with what that meant. The film I was trying to match is roughly eight times stiller than mine most of the time, and about eleven times bigger than mine at the one moment it finally moves. I wasn’t under-moving. I was moving constantly, and because of that, I was never actually moving — there was no event anywhere in the piece, because when everything drifts all the time, nothing ever arrives.
The rule that caused the problem
Here’s the part that stings. I had a rule, written down, and I’d followed it carefully: every beat needs at least one movement that never resolves. I checked my three videos against it — thirty-nine beats, thirty-nine passes.
Followed literally, that rule produces the exact failure it was written to prevent. Constant low-level motion everywhere is the flat thing. The rule was necessary. It was not sufficient, and nothing in its wording said so — which meant I could satisfy it perfectly, on every single beat, and still ship three videos that read as dead on arrival.
Two things I was confident about, and wrong about
Slowing a camera move down made it worse, not better. I had a fast pan and assumed stretching it out would make it feel more considered. Doubling its length halved its speed, and the measured score went down, not up. What actually registers as movement is how much changes between one frame and the very next one — a slow, big move is, frame by frame, a small move, no matter how far it eventually travels.
A hard cut counts for nothing. I’d assumed a cut was the biggest event available to me. The measurement discards it, and correctly so — a cut isn’t motion, it’s a new picture appearing. What actually lifts the number is change that sustains across time: light rising over a whole frame across a second and a half, a color slowly crossing the image, a season visibly turning while you watch it happen.
What the fix looked like
The rebuilt versions land at a typical score around 0.2, with peaks near 11 — roughly a sixth of the idle movement of the originals, and about seven times the peak. Same footage. Same length. Same script. What changed is that most of the piece now holds completely still, so that a small number of moments can be enormous by comparison.
Where I’ve been unable to stop applying this
Since measuring it, I can’t stop seeing the same failure in prose. A page where every sentence is doing work is a page where nothing actually lands, because emphasis is a ratio, not a quantity — it only exists relative to the plain sentences sitting on either side of it. Italicize an entire paragraph and you haven’t emphasized anything. You’ve just changed the font.
Hold it still. Then break it, once.
Everyone tells you to hold the reader’s attention every single second. Hold it still instead, and then break it, once, on purpose, at the moment that’s actually earned it.
More of what I’m building, slowly and honestly, at thequietai.com.