Most flat-looking AI videos have one thing in common: no clear camera instruction. Text-to-video models understand standard film language, so naming the camera move is one of the easiest ways to improve results. Here are 25 moves with copy-ready prompt examples.
How to write camera instructions
- One move per shot (or per beat in longer clips). Several moves at once confuse the model.
- Add speed: slow, smooth, steady, fast, sudden.
- Add direction: left, right, toward, away, around.
- Put the camera last in the sentence, after subject and action, so the model knows what to frame.
25 camera movements for AI video
1. Push-in (dolly in)
The camera moves toward the subject, building focus or tension.
Prompt: Slow push-in on a woman's face as she reads the letter.
2. Pull-out (dolly out)
The camera moves away, revealing context or isolation.
Prompt: Slow pull-out from a man on a bench to reveal an empty stadium.
3. Tracking shot
The camera follows a moving subject at the same speed.
Prompt: Side tracking shot following a runner along the beach at sunrise.
4. Truck left/right
The camera slides sideways, parallel to the scene.
Prompt: Smooth truck right along a row of market stalls.
5. Pan
The camera rotates left or right from a fixed point.
Prompt: Slow pan from the mountain range to the cabin.
6. Tilt
The camera rotates up or down from a fixed point.
Prompt: Tilt up from the boots to the face of the astronaut.
7. Orbit (arc shot)
The camera circles around the subject.
Prompt: 360-degree orbit around the sneaker on a pedestal.
8. Crane up
The camera rises vertically, often revealing a wide view.
Prompt: Crane up from the street café to the city skyline.
9. Crane down
The camera descends into the scene.
Prompt: Crane down from treetops to two hikers at a campfire.
10. Drone flyover
A high, smooth aerial move over a landscape.
Prompt: Drone flyover above turquoise coastline, slow forward motion.
11. FPV drone
Fast, agile first-person drone movement.
Prompt: FPV drone dives down a waterfall and skims the river.
12. Handheld
Slight natural shake for realism and urgency.
Prompt: Handheld shot following a reporter through a crowd.
13. Static / locked-off
No camera movement; the action carries the shot.
Prompt: Static wide shot of a storm rolling over the field.
14. Whip pan
A very fast pan that blurs between subjects.
Prompt: Whip pan from the chef to the flaming pan.
15. Rack focus
Focus shifts from foreground to background (or back).
Prompt: Rack focus from the raindrops on the window to the couple outside.
16. Dolly zoom
Camera moves while the lens zooms the opposite way; the background warps.
Prompt: Dolly zoom on the detective as he realizes the truth.
17. Top-down (bird's-eye)
The camera looks straight down.
Prompt: Top-down shot of hands plating a dish.
18. Low angle
The camera looks up, making the subject feel powerful.
Prompt: Low-angle shot of the skyscraper against moving clouds.
19. High angle
The camera looks down, making the subject feel small.
Prompt: High-angle shot of a child alone in a big library.
20. Dutch angle
The frame is tilted for unease.
Prompt: Dutch angle on the flickering hallway light.
21. Over-the-shoulder
Framed from behind one character, looking at another.
Prompt: Over-the-shoulder shot of the interview, focus on the guest.
22. POV
The camera sees through a character's eyes.
Prompt: POV walking through a narrow alley toward a glowing door.
23. Slow motion
Action plays slower than real time.
Prompt: Slow motion of the glass shattering on marble.
24. Time-lapse
Time speeds up over a static or slowly moving frame.
Prompt: Time-lapse of city traffic from dusk to night.
25. Hyperlapse
A time-lapse where the camera also travels.
Prompt: Hyperlapse moving through a busy train station.
Combining moves in longer clips
Models that generate 15–30 seconds (such as Kling 4.0, Seedance 2.5 and Wan 3.0) can handle a sequence of moves if you give each a time range: "0–5s: static wide shot. 5–12s: slow push-in. 12–20s: orbit left around the subject." See the Kling 4.0 prompt guide for examples.
Related guides
Try any of these moves in the TextVidAI generator.
FAQ
Do AI video models understand camera terms?
Yes. Current models respond to standard terms like push-in, orbit, tracking shot and crane up, especially when you add speed and direction.
Why does my camera move get ignored?
Usually because the prompt lists several moves, or the camera instruction is buried. Use one clear move and place it after the subject and action.
Last updated: October 9, 2026.
