Prepare a Multilingual Launch Film: Website Evidence, Native Labels and Review
Start from a public page or approved HTML snapshot, review language declarations and product copy, then hand off a portable brief for local video rendering. Declarations alone do not establish complete translations.
Fastest way
Ask your AI to do it
Paste this into the AI agent you already use:
Use Language Launch Film to prepare an announcement from my explicitly selected public page or approved HTML snapshot. Preserve declaration evidence and unsupported codes, review actual language support and replace starter product copy. Choose workspace, academy or commerce, inspect image rights/crops and check source before local encoding. Keep the draft unreviewed until confirmed; do not infer complete translation or use paid generation services.
Which part of Language Launch Film works before video setup?
Language Launch Film v1.2.1 makes multilingual app, course or collection announcements using workspace, academy or commerce stories. The individual package costs USD 4.99. Its site-first review form uses Python 3.10+ and the standard library before you install rendering dependencies. Local MP4 export uses the purchased renderer and pinned requirements; no translation, TTS or image-generation API is required.
These are three different visual compositions, not three theme colors. Choose workspace for a contained application screenshot, academy for course artwork or commerce for a collection image. Confirm whether you are announcing newly launched languages or describing existing support. A language selector or HTML declaration cannot establish that all screens and content are translated.
Inspect one authorized page and preserve the original declaration evidence
Read QUICKSTART.md and references/input.md. For your explicitly selected public site, run the first command. It fetches one anonymous HTTPS page with bounded same-host redirects, not an authenticated session or JavaScript-rendered language menu. If public DNS is unavailable or you have an approved snapshot, use the second command with the actual intended site URL. Do not disable the public-host checks.
| Observed markup | Handling and limitation |
|---|---|
| Root lang="en" and alternate hreflang="fr-CA" | Suggest en and normalized fr; preserve source hash |
| Body lang="ar" quotation or a /de link | Not evidence of supported Arabic or German product content |
| Unknown hreflang="xx" or x-default | Retain unsupported evidence; x-default is not a language |
| A completed discovery draft | reviewed:false; no movie rendered or translations verified |
python3 scripts/prepare-launch.py --url https://YOUR-PUBLIC-SITE/ --example workspace --output ./launch-brief
# Alternative for an approved local snapshot:
python3 scripts/prepare-launch.py --html snapshot.html --site https://YOUR-PUBLIC-SITE/ --example workspace --output ./snapshot-briefReview the suggested brand, native names and starter story
Open launch-brief/index.html. Inspect language-evidence.json and the copied source bytes before accepting the suggested product name and language checkboxes. Resolve unsupported codes and conflicting declarations manually. Under Customize the story, replace the headline, section, three cards, badges, visual title/detail and CTA with approved product copy. Starter wording is guidance, not text extracted and verified from your site.
An imported PNG/JPEG/WebP image is embedded in the saved brief; remote images are not automatically downloaded as verified logos. Review rights and the actual crop. Workspace contains the screenshot; academy/commerce center-crop artwork. Preserve fictional status for practice brands. Confirm language support only after your own review; the checkbox is a declaration, not a language-quality certification.
Inspect the current example output → — Compare the visible result with the exported files and checks described in this section.
Check the editable source before encoding the movie
Install requirements.txt in a virtual environment, then use --source-only for a reviewed downloaded brief. This still checks font coverage, shaping requirements and layout; it does not create an MP4. Use a fresh output directory. A standard-library discovery draft left reviewed:false must be rejected, not silently rendered.
python3 -m venv .venv
.venv/bin/python -m pip install -r requirements.txt
.venv/bin/python scripts/start.py --input edited-launch.json --source-only --output ./launch-sourceRender locally after confirmation and review the real crop
After the source check, omit --source-only and choose another new output folder. For an unchanged fictional starter, the command below creates one landscape film. The helper defaults to rendering; custom identity/copy flags require --confirm-languages. Do not use that flag to bypass an unresolved product-support claim.
- Arabic and Hindi require RAQM-enabled Pillow even for source-only layout checks. Use the bundled Dockerfile when native shaping is unavailable; its render can run with --network none. Docker must already be installed, and its first build downloads public dependencies. Never remove Arabic or Hindi merely to conceal unsupported shaping in a requested export.
- Check native spelling, direction, text fit, all requested crops, captions and actual audio. The 6–20 second duration and language count must leave readable per-language intervals. A successful full decode or glyph check does not replace native-reader review.
python3 scripts/start.py --example workspace --formats landscape --output ./first-language-filmKeep an edited brief distinct from its previously encoded film
Browser edits change the brief, not the current movie. Save edited-launch.json, retain any original local logo/music/voice assets it still references, and rebuild to refresh the actual MP4. Deliver editable JSON, licensed fonts/notices, storyboard, VTT, provenance, rendered files and SHA-256 manifest. Do not describe a checked language list as proof that the actual application is completely localized.
Audit real JSON/XLIFF catalogs separately from the announcement
Prepare source-backed product launch messages before publishing