REC 24 models · 6 points · 28 camera words · read 2026-09-25

How much film language each AI video model understands

A register of 24 AI video models and tools and what their own documentation lets you write about the camera: moves, shot sizes, several shots in one clip, start and end frames, negative prompts and motion settings, with 28 camera words traced to the docs that use them. As of 2026-09-25.

1The slates

One slate per model, in alphabetical order. The six cells along the bottom follow the order of every model page, MOV for camera moves, FRM for framing, CUT for cuts, FRS for frames, NEG for negative prompt and MOT for motion amount, and are lit where the maker's own docs cover the point. The foot of each slate counts the camera words its docs use.

2The camera words, by how many docs use them

Each bar is the share of the 24 models whose docs define the word, list it as a command or preset, or use it in an example prompt. The most widely used reach 10 models; the rarest, two.

3How many models cover each point

Fig. 1 Counted from the register. A point the docs skip, advise against or hide behind a sign-in is not counted.

Models whose docs cover each point, out of 24, read 2026-09-25.
Point	Covered	Not covered
How a camera move is asked for	21	3
Shot sizes and angles in the docs	12	12
Cuts inside one generation	7	17
Start and end frames	21	3
A negative prompt	6	18
A setting for how much the picture moves	6	18

Inclusion rule. The 24 models on the register; a point counts when the maker's docs describe it. Order. The order the points take on every model page.

4Why the docs, not the demos

A demo shows what a model did once. The docs say what a user can ask for: which words the model has been taught, which settings exist, which inputs it takes. For directing a scene, that is the part that decides whether a close-up or a dolly move can be requested at all or has to be hoped for.

Every reading here is quoted from the maker's own prompt guide, API reference or product page, and dated.

5What stands out

Nearly every model says how to ask for a camera move, and most take a start and end frame in some form. Fewer define shot sizes, fewer still say one generation can hold several shots, and only a handful document a negative prompt or a setting for how much the scene moves.

6Where to roll from

Start from a model if you already use one, from a camera word if you want to know which models understand it, or from a shot question when you have a scene to direct.

Video models — 24 AI video models read in their own docs…
Data — Every reading Text2Movie took from AI video model docs, in one CSV file
Camera controls — Four pages on the tools people search when a shot has to be directed…
Camera fields — Each camera point across all 24 models, with single-model pages where the docs say more
Camera words — Pan, dolly, orbit, close-up and more…
Model pairs — Two models' docs set side by side on the six camera points, with the differences quoted
Shot questions — Directing questions about AI video shots, answered with the models whose own docs cover them
Directing notes — Short notes on writing camera language for AI video models, grounded in their docs