Symphony Creative Studio reference clips must be under 10 seconds
TikTok's Symphony help page limits reference videos to under 10 seconds and single scenes. Cut your clip first with Sume's video trim, then upload it.

TikTok's Symphony Creative Studio help page says reference videos must be under 10 seconds (read 2026-10-02). If your best clip is longer, cut it to the moment you want to copy before you upload. Sume's video trim does that for $0.02 per job, according to its docs.
What the help page lists
The page describes three modes: text to video, image to video and reference to video. It says Dreamina Seedance 2.0 restricts some capabilities, including reference images or videos with real faces and celebrity likenesses (read 2026-10-02).
| Item | What TikTok says |
|---|---|
| Modes | Text, image and reference to video |
| Reference video | Under 10 seconds |
| Scene | Single scene only, no transitions |
| Layout | No split-screen or picture-in-picture |
| Real faces | Not allowed as references |
| Label | AI-generated label added to all exported videos |
Trim a reference clip with Sume
Sume's video trim takes one workspace clip plus a start and exactly one of end or duration, and returns a new MP4. The source is untouched. The clip must already live on media.sume.com, so import it first; a write needs an Idempotency-Key.
- Pick the 6 to 8 seconds that show one scene and one action.
- Trim with
startandduration, and readactual_start_secondsin the result. - Download the new MP4 and upload it as the Symphony reference.
- Check there is no split-screen layout or real face before you submit.
Limits
Trim only cuts time. It does not remove a real face, split a multi-scene clip into one scene or change layout. If the source has a split-screen design, crop to one panel with video filter, which supports crop operations and an unbilled check endpoint, or choose another clip.
Sources
Related posts
More in Use cases
- TikTok video_cover_timestamp_ms: pick the cover frame from a clip
TikTok takes a cover frame as a millisecond offset. Pull candidate stills with Sume video frames, choose one, and send its time as video_cover_timestamp_ms.
- Toast menu images: 750x450, 5 MB, crop-safe dishes
Toast recommends 750 x 450 px menu item images up to 5 MB that may be cropped. Generate a 3:2 plate with Sume, then crop to 5:3 with a short script.
- Translated with Meta AI label: how viewers turn Reels dubbing off
Meta labels every translated Reel Translated with Meta AI. A viewer picks Don't translate in the audio and language section of the three-dot menu to opt out.
- Trim a long video then caption the clip: two Sume jobs in order
Cut a moment from a long recording with video-trim, then burn captions on the new MP4. Why the order matters, what each job costs, and which URL goes where.
Written by Sume