1. Action-group classifier conflated object-touches and person-touches.
"man touches the red door with a small object" was being tagged as
forehead_touch because "touch" was in the forehead_touch needles set.
That made the realign pass yank Beat 16 from scene 451 (correct: man
painting red door, IV stand) over to scene 623 (woman/man in bed) —
a totally wrong shot at score 0.344.
Fix: removed generic "touch*" verbs from forehead_touch's needle set.
forehead_touch is now added in _semantic_action_groups() only when a
touch verb is paired with an explicit body-part target (forehead,
face, cheek, head, hand, ...) and not paired with an object target
(door, handle, brush, tool, lock, ...).
Effect on Beat 16 after `match --beat 16 --vision`:
scene 623 in=5476.28 score=0.344 -> scene 451 in=3912.48 score=0.626.
2. Cutter-report stills/clips were keyed by source-video mtime, so a
match-position change without a video change served stale frames from
the previous match. Dropped the mtime cache; both extractors now
render fresh every time. Slower (~minute per full regen) but correct.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- cli.py auto-regen now produces video clips on every match (no opt-in
flag). Best presentation comes first; speed cost (~minutes per match)
is accepted.
- output/cutter_stills/ and output/cutter_clips/ are no longer gitignored.
All 45 stills and 45 short MP4 previews are committed alongside the
CUTTER_REPORT.{md,html} so the remote repo always shows the current
state — even when the report files are inspected without running the
generator.
- Other output/ contents (FCPXML, EDL, debug folders, HTML report) stay
ignored.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Melbar
