explainer-kinetic-word-editlisted
Install: claude install-skill Safiullah-Rahu/claude-video-skills
# Kinetic word edit — six worlds, every word on screen, no hard cuts
Reverse-engineered frame by frame from one Instagram reel (**@themotionaddict_**, `DfMKeK6iPDN`, 37.1 s, 1:1, 311k likes;
`references/sources.md`). The source is a lyric edit of a commercial song with the artist's album art and photos: **none
of that is reused, and the song's words are not written anywhere in this skill**. What is kept is the edit grammar: the
scene chain, the way each word enters, the transitions, the beat-matched timing and the closing question. The words
come from an original spoken-word poem.
Sample: `the sample film (the render is not included in this repository)` — "Undertow"
(37.2 s, 9:16; `_nomusic.mp4` beside it; SCRIPT.md, SAMPLE.md, POST.md, out/qa/; the sample project is not included).
## 1. Signature traits
- **One continuous edit**: six visual worlds chained by focus pulls, whips, flashes, burns and crossfades. The analyzer
finds **0 hard cuts**; the pace comes from words landing on beats (one every 0.3–0.6 s in the type scenes).
- **Every sung/spoken word is on screen**, each time in a different material: tone-on-tone title in the water, glowing
3D letters, a highlighter on typed paper, fire, marker on a crumpled note, white type around a question mark.
- **Ghost-then-light**: the next word sits grey and dim ~0.3–0.4 s before its syllable, then lights (a read-ahead).
- **A stretched vowel**: the last word of the type section grows letters (ours: SLOOOOW) while the camer