Localising vertical drama into 12 languages without re-shooting
What survives translation in a 60-second episode, what has to be re-recorded, how subtitle timing differs per language, and why dubbing a vertical drama is a casting decision rather than a translation decision.
This guide is published in English. Titles and summaries are localised; the full text remains in the original language.
Vertical drama travels. A revenge plot with a clear reversal works in Jakarta, São Paulo and Cairo for the same reason it works in Seoul: the emotion is carried by faces and cuts, not by wordplay. But “travels well” is not the same as “can be translated mechanically”.
What survives, what breaks
Survives: reversals, status conflicts, family secrets, betrayal, contract marriages, revenge reveals. Anything driven by a visible power shift.
Breaks: puns, dialect jokes, honorific subtleties, legal specifics tied to one jurisdiction, and anything where the joke is the wording rather than the situation.
The practical test: cover the subtitles and watch. If you can still follow the story, it will localise. If you cannot, no translation will save it.
Subtitles are a timing problem before they are a language problem
A 60-second episode has room for roughly 90–130 characters of subtitle per language at readable speed — and that number is not constant across languages.
| Language | Expansion vs English | Practical effect |
|---|---|---|
| German | +15–25% | Lines must be shorter in the English script, not squeezed in the subtitle |
| Spanish | +15–20% | Same |
| Russian | +10–15% | Same |
| Japanese | −10% (but 2 lines needed) | Vertical space, not horizontal |
| Korean | −5% | Fits well |
| Arabic | +10%, RTL | Right-to-left rendering, mirrored punctuation |
| Hindi | +5–10% | Devanagari needs larger type at the same size |
| Chinese (Simplified) | −40% | Very compact; avoid over-tightening the English |
The consequence for writers: write short lines in the source language. If your English line is eleven words, the German subtitle will wrap to three rows on a phone and the viewer will lose the beat. Nine words is a ceiling, six is comfortable.
Dubbing is casting, not translation
A cloned voice does not automatically speak a second language with the same timbre, rhythm or age. Two honest options:
- Re-cast per language. Pick a voice per language that matches the character’s age, energy and register. Costs more, sounds native.
- Keep the original voice, subtitle only. Cheapest, and for many markets the correct choice — vertical drama audiences are subtitle-literate.
What does not work: machine-translating the audio and keeping the lip sync from the English pass. The mouth shapes belong to English phonemes. If you dub, re-run lip sync against the new audio, and budget 18 credits per dialogue shot for it.
The localisation checklist we use
- Lock picture before localising. Any re-edit invalidates every subtitle file.
- Translate from the shooting script, not the storyboard notes.
- Cap subtitle lines at 42 characters for Latin scripts, 20 for CJK, and test on a 6-inch screen.
- Check RTL rendering for Arabic — punctuation, ellipses and numbers all mirror differently.
- Keep character names consistent across languages; do not localise names unless the client asks.
- Re-run lip sync for any dubbed language, per dialogue shot.
- Label the export as synthetic media in the metadata of every language version.
Cost per language
Subtitles are 6 credits per episode per language. Twelve languages is 72 credits on top of a 600-credit episode — about 12%. Dubbing is the expensive path: 18 credits per dialogue shot, so eight dialogue shots across twelve languages is 1,728 credits, nearly three episodes of budget.
That ratio is why our default recommendation is subtitles in all 12, dubbing only in the two or three markets where you have distribution. Localise wide, dub narrow.