Camera prompts for AI video: a cheat sheet

·2 min read·Filmmaking

Most AI video prompts describe what's in the frame and forget who is holding the camera. Yet the camera is what separates “AI clip” from “shot”. Here is the vocabulary that models respond to, with ready-to-paste templates.

Generated in Actoria · Seedance 1.5 Pro · 15 sOpen the video page

Why camera words beat adjectives

Video models have seen millions of hours of film, and film has a grammar. “Slow push-in on her face” maps onto thousands of examples; “cinematic, dramatic, masterpiece” maps onto nothing in particular. Use the film term and you borrow the whole convention: framing, speed, where the eye goes.

The cheat sheet

SayWhat you getBest for
static wide shotlocked-off camera, full sceneestablishing a place, groups
slow push-incamera creeps toward the subjectemotion, reveals, endings
pull-back / dolly outcamera retreats, reveals contextpunchlines, scale
pan left / rightcamera turns on its axisfollowing a walk, landscapes
tracking / follow shotcamera moves with the subjectwalks, chases, runways
handheldslight shake, documentary feeltension, UGC-style ads
low anglecamera below eye line, subject looks biggerpower, heroes
high angle / top-downcamera above the subjectvulnerability, tables, food
half orbitcamera arcs around the subjectproducts, character reveals

Templates

  1. [Subject] [action], [camera move], [light]. — “She reads the letter at the window, slow push-in, soft morning light.”
  2. [Camera move] across [place], [time / weather]. — “Slow pan across the harbour at dusk, light rain.”
  3. Handheld follow behind [subject] as [action]. — “Handheld follow behind him as he pushes through the crowd.”

Motion intensity

In Actoria the Motion selector (low / medium / high) controls how much the scene moves overall; the camera move comes from the prompt. Low motion plus a slow push-in gives you the quiet, expensive-looking shot. High motion plus handheld gives you chaos — great for one beat, tiring for five.

Mistakes that break shots

01
Compose
Build the still with space for the move.
02
Name the move
One camera term per scene, plus light.
03
Extend
The second move lives in the next scene, starting from the last frame.

For the full director's workflow — light, palette, continuity across cuts — see AI cinematic video; for stitching scenes into a film, the scene-by-scene guide.

Frequently asked questions

Do models understand lens terms like 35mm or anamorphic?

Partially — they bias the look, but camera-movement terms are far more reliable than lens terms.

Can I control speed?

Use words — slow, gentle, rapid — and lower the Motion setting for calmer shots.

What about vertical video?

Same vocabulary; pick the 9:16 format when creating the project and compose the still for it.

Direct your first shot
Compose the still, name the move, watch it play.
Create your first video

More from the blog

All articles