A segment is not a shot, and the gap costs money
A segment is what one generation call returns. A shot is what an editor counts. Several models now cut inside a single call, so one segment can contain three shots, and an episode priced per shot will be priced wrong. As of 2026-09-12.
| Model | Seconds per call | Cuts inside one call |
|---|---|---|
| Kling VIDEO 3.0 | 3 to 15 | Up to six shots in Multi-Shot |
| MiniMax H3 | 4 to 15, H3-Max 5 to 15 | Multi-shot prompts on the model page |
| Seedance 2.0 | 4 to 15 | Storyboard prompts with transitions, as the vendor states |
| Wan 3.0 | 2 to 30 | Multi-shot storyboard with timestamps, 4 to 6 s per shot |
Inclusion rule. Models whose vendor documents both a per-call length and whether cuts can happen inside one call. A model that publishes only a length is not given a row. Order. Alphabetical by model name.
1Why the distinction is expensive
Generation is billed per second of returned video, not per shot. If an episode is planned as forty shots and the model cuts three shots inside one call, the production pays for perhaps fifteen calls rather than forty — or, planned the other way round, budgets fifteen and needs forty.
The error compounds across a season, and it is invisible until the first invoice, because a shot list and a rate card use the same word for different things.



2Cuts inside a call are not free
A model that plans its own cuts decides where they fall. That is fine when the cut is a beat the edit wanted and expensive when it is not, because fixing it means regenerating the whole segment rather than trimming a clip.
Where the vendor documents an explicit multi-shot syntax, the cuts become something the shot list can specify. Where it does not, they are a property of the model on that day.
3The ceiling decides the grammar
A model returning at most eight seconds cannot hold a shot that needs twelve, so the shot list either shortens the beat or accepts a join in the middle of it. Neither is free, and the choice belongs to the writer rather than the editor.
Ceilings differ by a factor of ten across the models here, which means the same scene is written differently depending on which one will render it.
4What to write in a shot list
Two numbers per row: how long the picture runs, and whether it is one continuous view or several. The first prices the call; the second decides whether the call can deliver it at all.
Productions that keep both survive a change of model. Productions that keep only the shot count have to re-plan every time a ceiling or a multi-shot syntax changes.
5Where these figures come from
A definition kept for this register: the term, and what vendors state about it, each figure read on 2026-09-12. The tools themselves are on the register. Related: Character sheet, Reference image.
6How a serial stays consistent
Serial work like “Segment and shot” depends on continuity; SceneMixer's guides describe how it keeps it:
SceneMixer separates the script language from the production language: the output (episode outlines, storyboards, on-screen text) follows your interface language, and target-region localization adapts character names, casting, and scenery for Western, East Asian, Southeast Asian, or Latin American audiences.
SceneMixer, Turn a Novel into a Video Series with AI — “Can I turn a Chinese web novel into an English-language drama?” (read 2026-10-13)
Typically 60–120 seconds, always 9:16 vertical, ending on a hook. SceneMixer's episode outlines are generated with that pacing, and setting a project to 9:16 makes every segment vertical (the default is 16:9).
SceneMixer, How to Make ReelShort-Style Dramas with AI — “How long is a ReelShort-style episode?” (read 2026-10-13)