Two type voices cut hard on every spoken word, one or two words at a time. Italic Instrument Serif for the claim, bold Inter for the instruction, and a cyan-gradient pop pair reserved for the payoff phrase. All frames below are generated by scripts/build-captions.mjs from examples/apple-skill.plan.txt over cleaned stills of the reference reel.







using > this one — italic lead word locked over the bold payoff; payoff words pop in one by one on their spoken onset.
into a > single skill — white→cyan gradient, drop shadow, dark blurred oval plate behind the pair.
comment > "Apple" + — the CTA keyword; + holds it to the end of the shot.







| ink | #111111 on the white page |
| paper | #FFFFFF on the black page / over footage (+ shadow) |
| payoff gradient | #FFFFFF → #C7F2FF → #2ACBF7 · on-brand swap --accent "#9FB0EC" --accent-mid "#E3E8FF" |
| plate | rgba(0,0,0,.42), 620×280 oval, blur 34px |
| serif | Instrument Serif Italic 108px (caps or lower), tracking -0.01em |
| sans | Inter 700 78px gap / 66px chin, tracking -0.025em |
| pop | lead Instrument Serif Italic 104px · payoff Inter 800 94px · overlap -36px · pop 0.18s power3.out scale .84→1 |
| slots | gap 930px (48%) · chin 1400px (73%) · chin-pop 1410px · y=<px> override |