HyperFrames faceless explainer prompt — narration-ready cards

<direction> A 9-second faceless explainer for HOW HABITS ACTUALLY FORM, 1280x720 landscape — no presenter, no stock faces, just cards a narration track can sit on. Three acts: three numbered headline cards (CUE / LOOP / LOCK), each with a headline and one body line; then a three-box diagram draws itself and connects; then ONE SMALL REPEAT, EVERY DAY. lands as the closer. Every act gets a clean full-screen beat so the voiceover never fights the visuals. Banned: presenter or avatar renders, stock-photo people, glows, gradients, walls of paragraph text, more than one accent. </direction> <structure> Act 1 (from 0.3s): three cards stagger in, each headline + body. Act 2 (from ~3.4s): diagram boxes draw, arrows connect them in reading order. Act 3 (from ~6.3s): diagram yields, closing line rises, small print states the card timings are narration beats. Hold to 9s. </structure> <build> One HyperFrames composition, seek-driven GSAP. Diagram connectors are SVG lines drawn with stroke-dashoffset measured via getTotalLength. Cards are fixed-height boxes — size them for the two-line headline case, not the one-line case. </build> <gotchas> First snapshot: the third card's two-line headline pushed its body text into the next card. Fixed by measuring the wrapped headline (two lines at the chosen size) and moving the body down + widening all three cards. Always lay out for the wrapped state. </gotchas> Before you finish: step through the timeline at five points and fix anything that overlaps, clips the frame, or drifts off the beat grid.