1. What this lesson is for

This is the capstone. You take one project from brief to published video - twice: once in English and once in Polish - using everything from the course: cutting, audio, captions, colour, AI voiceover, packaging, templates and a proper review cycle.

Your deliverable is the module's full delivery package: two published videos, versioned files, a revision log and bilingual final exports. Treat yourself as the client: predictable process first, flashy edits second.

2. Planning two versions from one core timeline

Do not edit two separate videos. Build one core timeline with everything that is language-neutral, then branch it into EN and PL only for the parts that carry language.

• Lock the picture first. Only after the core cut is approved do you duplicate the project into _EN and _PL copies.
• Plan for length differences. Polish narration often runs longer than English for the same content. Leave breathing room in B-roll sections so the longer language fits without re-cutting the picture.
• Keep text off the core layer. Any on-screen words belong to a branch, or you will be translating inside the locked cut.

CAPSTONE PROJECT STRUCTURE
/CAPSTONE-2026/
  00_brief/          brief.txt, script_EN.txt, script_PL.txt
  01_source/         footage, music, sfx
  02_vo/             vo_EN_v1.wav, vo_PL_v1.wav, ...
  03_projects/       CAP_core_v3, CAP_v3_EN, CAP_v3_PL
  04_exports/        review and master exports
  05_delivery/       masters, .srt, thumbnails, metadata.txt
  06_logs/           revision-log.txt, qa-matrix.txt, retro.txt

3. Using ElevenLabs voiceovers in both languages

Generate each voiceover from its own approved script, not by translating on the fly. Keep the voice settings identical between takes of the same language so re-generated lines match the rest.

1. Finalize script_EN, then write script_PL as a natural adaptation, not a word-for-word translation. Have a native speaker read it if you can.
2. Generate each language in ElevenLabs with a voice that suits it; note the voice and settings in your log so you can regenerate single lines later.
3. Export at the quality you used earlier in the course and name files by language and version.
4. Place each VO in its branch and align it to picture beats. Adjust pauses in the audio first; touch the picture only if a beat truly cannot land.
5. Listen to both versions end to end at 1x, checking names, numbers and brand terms.

Run the two branches on one control board so neither language falls behind:

4. Final quality checklist before publishing

Check each language against the same criteria and record evidence - a timestamp, a screenshot, or a note - not just a tick.

DELIVERY SPEC - CAPSTONE
Video:     H.264 MP4, 1920x1080 (16:9) + 1080x1920 (9:16), 30 fps
Audio:     AAC stereo, dialogue level consistent between EN and PL
Captions:  CAP_v3_EN.srt, CAP_v3_PL.srt (UTF-8, Polish characters checked)
Files:     CAP_v3_MASTER_EN.mp4, CAP_v3_MASTER_PL.mp4, thumb_EN.png, thumb_PL.png
Metadata:  title, description, tags, chapters - written per language
Logs:      revision-log.txt, qa-matrix.txt, retro.txt

5. Publish sequence

• Upload as private or unlisted first and watch the processed file on a phone.
• Metadata check: correct language set, captions attached to the right language, thumbnail matches the video language, links work.
• Publish both versions in the same planned window and note the exact time.

6. Post-publication review and improvement notes

ANALYTICS SNAPSHOT        EN (48 h)   PL (48 h)   EN (7 d)   PL (7 d)
Views                     ____        ____        ____       ____
CTR (impressions)         ____        ____        ____       ____
Avg view duration         ____        ____        ____       ____
Retention at 30 s         ____        ____        ____       ____
Biggest drop-off (MM:SS)  ____        ____        ____       ____

RETRO
Key wins:              ______________________________
Quality bottlenecks:   ______________________________
Next cycle, change:    1. ________  2. ________  3. ________

7. Capstone rubric

CAPSTONE RUBRIC (score 0-2 each, pass = 16/20 with no zero)
1. Core timeline reused; EN and PL differ only in language layers
2. Cutting and pacing: no dead air, clear structure
3. Audio: clean dialogue, consistent loudness in both versions
4. Voiceovers synchronized to picture beats in both languages
5. Captions accurate, readable on phone, correct characters
6. Visual quality: colour, framing, graphics consistent
7. Packaging: title, thumbnail, description per language
8. Versioning and file naming consistent from v1 to MASTER
9. Revision log and written approval present
10. Published + analytics snapshot + retro completed

8. Exercise

1. Write a one-page brief and both scripts (EN and PL) for a 45-90 second video.
2. Build and lock the core timeline, then branch it into _EN and _PL projects.
3. Generate both voiceovers in ElevenLabs, sync them, and add captions and titles per language. Track progress on the control board.
4. Run one review round using the feedback form and revision log from lesson 7.2.
5. Fill in the QA matrix, fix every Fail, and export using the delivery spec.
6. Publish both videos following the publish sequence and take the 48-hour snapshot.
7. Score yourself with the rubric and write the retro: key wins, quality bottlenecks, and three concrete improvements for your next bilingual release.

Further reading

• CapCut Help Center
• ElevenLabs documentation
• YouTube Help - Add subtitles and captions
• YouTube Help - Using YouTube Analytics