Iteration 1 — v4_A — first sticky-song pass (R&B / soul)
below_threshold
What changed
Replaced earlier piano-and-narration treatment with a soul-R&B sticky song. Lyrics rewritten under Trellison untethered-truth policy: no institution name-checks in the body; receipts on the lit-review page.
Why the prior version fell short
Earlier pre-v4 IMLS literacy videos were image-loop + voiceover with no melodic hook — easy to make, easy to forget. Set baseline = Napoleon (animated data-viz, 2.5–5s cut cadence, multi-layer visual interest).
Sticky song: The library effect grows where you cant see / Talk, sing, read, write, play / That is how words stay
Below threshold for standalone arsenal — visual layer is image montage only, no animated data-viz overlays. Shipping here as iteration record, not as deliverable.
Iteration 2 — v4_B — alt arrangement (rock / country tilt)
below_threshold
What changed
Same lyrics, different musical arrangement. Used as A/B counterpart so audiences can pick the voice they respond to — same evidence base, two musical entry points.
Why the prior version fell short
Same visual gap as v4_A — montage only.
Sticky song: Same as v4_A
Iteration 3 — v5 (planned) — overlay-augmented
planned
What changed
Phase 2 render driver lands transparent-alpha WebM overlays per scene (counter / ring / bar_grow / sparkline / lower-third callouts) composited over the v4 visual track. Adds the animated data-viz layer that pushes the visual past the Napoleon baseline.
Why the prior version fell short
v4 stripped overlays during ffmpeg compose because filter_complex was too brittle. v5 uses per-scene isolated MP4 + concat demuxer + overlay clip compositing.
In progress under task #33 (Phase 2 render driver). Once accepted, v5 ships standalone on the arsenal hub and posts can re-add it via linked_artifacts.music_video_v5.