6 Commits

Author SHA1 Message Date
2e3ad6ab88 feat: add brand-aware motion presets (brand/presets/) + apply-preset.mjs
Five named structural presets (bold, calm, editorial, energetic, minimal) that
set the pacing/ground/exit knobs build.py already reads with a default (brand
mode/radius, film gap, music bpm/drop scale, per-beat-type ground/exit/hold)
without ever touching brand identity (colors, fonts, logo, name, tagline,
domain, url, description stay exactly what brand-from-url.mjs extracted).

tools/apply-preset.mjs applies a preset onto brand/brand.json + film/film.json:
a beat's own explicit field always wins unless --force, writes timestamped
backups before touching either file, and stamps a `_preset` block into both so
a later diff explains why gap/mode/radius changed. --list and --dry-run
verified against the repo's own Acme example (bold/minimal dry-run output
confirmed no writes and correct skip-when-hand-set behavior).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-10-28 02:47:33 +03:00
aaf2dfe31c feat: add multi-language render pass (tools/localize.mjs)
Translates film/film.json's lines and on-screen beat text, plus every
film/hooks/<hook>/hook.json, into one or more target languages via the free
unofficial Google Translate endpoint (no API key, cached by source-text hash
in tools/.translate-cache.json so re-runs after a copy edit only pay for what
changed). Writes a fully renderable project under film/i18n/<lang>/.

The hard part: film.json's beats carry literal English word anchors
(until/push_at/merge_at/flip_at/chip.at/cursor.at) that build.py's word()
resolves by position inside the matching spoken line. A naive translation
breaks every one of them. remapAnchor() re-anchors each key to the
translated line's word at the same proportional position, preferring a
semantically-matched emphasised word from the already-translated on-screen
phrase when one is available.

tools/render-lang.mjs (+ film/render-lang.sh thin wrapper) stages a
localized film.json and hooks/ over the live ones, runs the real
film/render.sh pipeline unmodified, copies results to renders/i18n/<lang>/,
then always restores the original - the same stage/restore pattern
tools/batch.mjs already uses.

Localized output points voice_dir at a sibling directory that never exists,
so build.py's and audio.py's own no-take fallback (silent animatic, timed
from the translated text itself) runs - without this, both scripts load the
*English* vo/body/lines.json and either crash on a translated anchor word
that was recorded in English, or time foreign on-screen text against English
speech. There is no recorded take in any language but English; a voiced
render needs a real foreign VO and a voice_dir pointed at it.

Ships with es/fr/ja examples (both hooks, Acme) generated and checked in,
same convention as the other example assets. Verified end-to-end: a real
`node tools/render-lang.mjs es type-only` produces a real 37s mp4
(renders/i18n/es/type-only.mp4), and film/film.json + film/hooks/ are
restored byte-identical afterward. Also fixes a gap from round 1: root
renders/ (used by batch.mjs and now render-lang.mjs) was never gitignored.
2026-10-23 22:32:42 +03:00
397f95f512 feat: add launch-import skill to convert competitor script exports
tools/import-script.mjs parses a flat, single-scene script export in
the shape shipvideo/Waymark/InVideo-style tools produce (one ad-hoc
scene per line, no beat structure) and maps it onto this kit's 6-beat
film.json grammar (name/hero/product/blocks/switch/end), ready to
render on the free path immediately. skills/launch-import/SKILL.md
documents the accepted input shapes and the beat-mapping heuristic for
an agent driving the conversion.

Gives readers who arrive with someone else's export a concrete
conversion path instead of a blank film.json.
2026-10-23 15:14:56 +03:00
bce5687f53 feat: add automated screenshot capture tool
tools/capture-screens.mjs drives headless Chrome to every route in a
JSON routes file, waits for real content via a named selector (never a
fixed timeout, so no loading skeletons get captured), and writes
consistently-cropped PNGs straight into film/assets/screens/.

Removes the one fully-manual step this kit's own README calls out as
the thing that "carries the film": bring your own 3-4 screenshots.
2026-10-23 15:14:49 +03:00
f47db146eb feat: add batch/portfolio render command for multi-product manifests
tools/batch.mjs (and the film/render-all.sh wrapper) accepts a JSON
manifest of N products, each with a brand source (URL or existing
brand kit), an optional screenshots dir and a film.json body, and
renders every one in a single pass: stage brand + screenshots +
film.json into this repo's own film/ and brand/, run the real
render.sh pipeline, copy the output into renders/<product>/<hook>*.mp4,
then restore this repo's example state before moving to the next
product (always, even on failure). Writes a combined
renders/batch-report.json summary.

Turns the ad-hoc by-hand workflow (swap film.json, render, repeat,
already used across several products) into one documented, reusable
command.
2026-10-23 15:14:41 +03:00
Tim Scheuer
bec0642ca5 Launch Video Kit: website to brand kit, the hook plus body film template, 10 launch skills, free audio path and the Acme example
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-09-27 12:14:37 +02:00