15-Second or 30-Second Segments

Open a finished episode and read the lengths on the timeline cards.
Where it is
Step 3, below the episode cards: Segment length — Standard 15s / Long take 30s. The hint beside it says the limit: 30s segments can only be rendered by models supporting 30s. Episodes you have already parsed keep the length they were cut to; the setting applies to the next Parse Storyboard or Re-parse. An episode cut into long takes carries a Long take 30s badge on its card.
Can I set a default for new storyboards?
Yes. In SceneMixer’s Settings › Model Preferences, Default segment length sets 15 or 30 seconds for every storyboard you generate from then on: Max length of each video segment when storyboards are generated; you can still change it per episode on the storyboard page. If your default video model cannot render 30 seconds, the setting says so; the Wan 3.0, Seedance 2.5 and Kling 4.0 tiers can, Seedance 2.0 and MiniMax stop at 15. The episode-level choice below the cards still wins for that episode.
What a segment is
One generation: a few shots that are made together and land as one clip. The shots inside it add up to its length. An episode is a row of segments on the timeline; the composite stitches them. Shot seconds.

Standard 15s
Segments of up to fifteen seconds: three to five shots each, a beat of a scene. More segments per episode, each cheaper on its own, each independently regenerable. If one comes out wrong you redo fifteen seconds, not thirty. Every video tier can render it. This is the default.
Long take 30s
Segments of up to thirty seconds: a longer stretch of a scene in one generation, with the camera cutting inside it. Good for a conversation that should not be interrupted by a generation boundary, or an action beat that needs the room. Fewer segments per episode. On SceneMixer, only the tiers that support 30-second clips can render them — the Wan 3.0 and Seedance 2.5 families — and a 30-second segment on a 15-second tier is refused as over cap until you switch tier. Tiers.
Price
Video is priced per second of segment, so an episode costs the same whether it is cut into fifteen- or thirty-second pieces. What changes is the size of each regeneration: redoing a long take costs twice a standard segment.
Changing your mind
Switch the setting and Re-parse the episode: it is cut again at the new length, and its existing segments and videos are discarded. Do this before you shoot, not after. What re-parsing loses. Within an episode you can also split one segment by hand. Inserting a segment.
| Standard 15s | Long take 30s | |
|---|---|---|
| Segment length | Up to 15 seconds | Up to 30 seconds |
| Segments per episode | More | Fewer |
| Cost of redoing one | Smaller | Larger |
| Tiers that can render it | All | Wan 3.0 and Seedance 2.5 families |
| Best for | Fine control, drafts | Uninterrupted conversations, longer beats |
FAQ
Which should I use?
Standard 15s unless you have a reason: it is the default, every tier renders it, and mistakes are cheaper to redo.
Does 30s cost more?
Not per second. An episode costs the same; each regeneration is larger.
My 30s segment says over cap.
The tier you selected in SceneMixer renders up to 15 seconds. Pick a tier from the Wan 3.0 or Seedance 2.5 families.
Can I mix lengths in one episode?
An episode is cut at one length, but you can split a segment by hand afterwards.
Does changing the setting change episodes I already parsed?
No. They keep their length until you Re-parse them. Re-parsing replaces that episode's storyboard and its videos are discarded.
I picked Long take 30s. Why is a segment only a few seconds?
Thirty seconds is the most a segment can hold, not a length every segment is filled to. The story decides where segments end; raise the seconds of a segment's shots in the editor to lengthen it. Set 30 seconds but got short segments.
Start standard, go long when the scene asks
Parse an episode at 15s, shoot a segment, and see if the beats feel right.
Updated November 15, 2026