How each model expects a shot to be written down
A storyboard script is the text a model consumes to produce a shot. Every vendor publishing a format has chosen a different one, so a shot list written for one model is not a shot list for another. As of 2026-09-12.
| Model | How shots are expressed | Length per generation |
|---|---|---|
| Kling VIDEO 3.0 | Multi-Shot plans cuts itself; Custom Multi-Shot takes them | 3 to 15 s |
| MiniMax H3 | Multi-shot text in one prompt | 7,000 characters per text item |
| Sora 2 | Separate shot blocks, one camera move each | 4, 8 or 12 s per clip |
| Veo 3.1 | Timestamped shots inside one clip | 4, 6 or 8 s per clip |
Inclusion rule. Models whose vendor documents how to express more than one shot. A model taking only a free-text prompt is not given a row. Order. Alphabetical by model name.
1Four formats, one production
Timestamps, separate blocks and single-prompt multi-shot text are three different ways of saying the same thing, and the conversion between them is not mechanical. A shot list has to be written for the model it will run on, or written in one plain house form and converted deliberately.
Productions that keep one house shot list survive a model change. Productions that write straight into a vendor's syntax re-author the season.
2Length limits shape the writing, not just the budget
A format allowing four, eight or twelve seconds cannot express a nine-second beat. The shot list has to be written against the ceiling from the start, which is a writing constraint rather than a technical one.
3Dialogue is handled differently again
One vendor takes a dialogue block below the prose, another expects speech in quotation marks, another takes reference audio. Where a shot list carries dialogue, that part converts least cleanly of all.
4Camera language does not transfer
One vendor takes a named camera move per block, another expects the move described in prose, another infers it. A shot list written in one vocabulary produces different framing in another, even when every other word is identical.
Keeping the intent in plain language and translating into the vendor's vocabulary at generation time is more durable than writing straight into whichever syntax is current.
5The script is the artefact that has to survive
Models change, tiers are retired and prompts stop behaving. What a production keeps is the shot list, so it has to be written to outlive the tool it was first rendered on.
That means recording what the shot should contain rather than what incantation produced it once. A shot list written that way can be re-rendered on a model that did not exist when it was written, which is the only form of future-proofing available here.
6Where 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: Scene plate, Vertical micro-drama.
7Continuity in practice
Serial work like “Storyboard script” depends on continuity; SceneMixer's guides describe how it keeps it:
SceneMixer generates each character once as a full-body reference sheet — 9:16, plain background, directional key light, identity features written before any technical instruction — and then attaches crops of that sheet to every shot the character appears in.
SceneMixer, AI Character Sheet Generator (read 2026-10-13)
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)