One prompt vs the harness (the 90/10 rule)

"Made with one prompt" is half true, and the half that is missing is the whole course. A one-liner like make a dynamic 15-second motion graphics showreel, go all out really can produce a watchable clip — because "showreel" is a genre the model already knows (fast cuts, a new technique each shot, best work first). It is a great way to confirm your rendering setup works.

But a one-liner tests your engine, never your idea, because it does not contain one. The pieces that actually land add five things the prompt alone cannot, and they are the five parts you will build:

1. A reference — without one, Opus defaults to the house style of AI video: centered text on a gradient, everything fading in, labels in the corners. (Module 3)
2. A render engine it controls — its own seek(t) renderer, or a framework like Remotion / HyperFrames. (Modules 1 and 4)
3. Real motion — closed-form springs, so movement has mass instead of a canned easing curve. (Module 4)
4. Sound on a grid — a measured or synthesized soundtrack with every cut and click on the beat. (Module 4)
5. A critique loop — Opus 5.5 reads images, so it can look at the frames it rendered, score them, and fix the worst ones. (Module 5)

Call it the 90/10 rule: the prompt is about 10% of a good video; the harness is the other 90%. Internalize it now and you will stop being surprised that your one-prompt clip looked "mid" while someone else's looked like an agency made it. Theirs had the other 90%.

Two practical dials before we build. Effort: Opus 5.5 runs at low, medium, high, xhigh, and max, and defaults to medium. Use medium for small fixes and re-renders; use xhigh or max for a new film where the first three seconds carry a launch. Cost: Opus 5.5 is $4 per million input tokens and $20 per million output (cache reads are $0.20), roughly 40% cheaper to run than Opus 5 — which matters, because a polished piece is dozens of render-and-critique passes, not one.