English and Korean UGC avatar ad: two jobs, Hangul captions, priced

One Avatar 1.0 job speaks one script. For English and Korean versions run two jobs; inline captions add no caption fee, but Korean needs a Hangul style.

5 min readSume
All posts

How do you make the same UGC ad in English and Korean with an avatar? Run two Avatar 1.0 jobs, one per script, and turn on inline captions in each; the caption stage is not a separate billed job. For a 20-second ad at the plus tier that is $4.90 per language, $9.80 for the pair.

Instagram made two-language captions a headline feature this month: its Edits bilingual captions translate a clip's captions into a chosen second language, across 15 languages. That is an in-app editing layer. If your ad ships as a rendered file that has to look the same on every placement, you burn the captions in at render time instead.

One script per job

The avatar request takes a script or ordered video_inputs, not both, and the spoken text decides what the captions say. There is no 'second language' switch on a single render, so the clean pattern is a pair of requests that differ only in the script and the caption style.

Give each request its own Idempotency-Key, for example ad-142-en-v1 and ad-142-ko-v1. A replay with the same key and body returns the original job instead of starting another one.

The caption rule that bites Korean

Inline captions accept the same settings as standalone video captions: style, optional font, a language hint and script_text. Styles include slam (the default), punch, tiktok-green, korean-ad and the Hangul identities weight-shift, black-outline, highlight, pill-karaoke, clip-wipe and editorial-emphasis.

If a Korean script is paired with slam, punch or tiktok-green, Sume rejects the request with 400 caption_hangul_text_latin_style. Sume does not swap the style for you, because those faces render Hangul as tofu. For the Korean job, choose korean-ad or one of the Hangul identities.

{
  "captions": { "enabled": true, "style": "korean-ad", "language": "ko" }
}

Cost for the pair

Both jobs bill per second of output at the tier rate. Caption failure is a soft failure: the avatar job can still succeed with a clean video_url and captions.status=failed, so you keep what you paid for and can re-caption the clean file with a standalone captions job.

20-second ad per language, no product image (rates read 2026-10-07)
TierRate per secondEnglish jobKorean jobPair
standard$0.184$3.68$3.68$7.36
plus (default)$0.245$4.90$4.90$9.80
max$0.55$11.00$11.00$22.00

When Edits is enough

If you only publish natively from Edits, its automatic second-language captions save you the burn-in step. Use the rendered route when the same file goes to paid placements, a landing page, or any surface where the platform's caption layer is not available.

Planning the pair

Treat the two language versions as a pair from the start. Write the Korean script as its own text, not a literal translation, and choose a Hangul caption style for it; the English job uses a Latin style such as tiktok-green. Send the same avatar_handle in both so the face matches.

  • Same avatar_handle, two scripts, two jobs.
  • Latin style for English, Hangul style for Korean.
  • Inline captions are not billed separately.

Publishing the pair

Upload each version as its own video and label each one as generated content where the platform asks. If the audience is bilingual, compare the cost of two Avatar jobs with a single job plus translated captions in your editor; the second path is cheaper but the voice stays in one language. Pick based on whether the audience needs to hear the second language.

Sources

Related posts

More in Sume Avatar 1.0

All Sume Avatar 1.0 posts

Written by Sume