Veo Camera Movement Prompts: Dolly, Orbit, Crane, and FPV Explained

Veo responds to real cinematography vocabulary — dolly, orbit, crane, pan, tilt, FPV — far more reliably than vague direction like “move the camera around.” Use exactly one named move per shot, state it plainly, and pair it with a framing (close-up, wide, medium) for the most predictable result. This guide covers the moves that actually work, what each one is for, and what happens when you try to combine them.

Why film vocabulary works better than plain description

Veo was trained on a huge volume of real film and video footage, which comes labeled and discussed using standard cinematography terms. “Dolly-in” maps to an enormous, consistent pattern in that training data; “the camera slowly moves closer in an interesting way” doesn’t map to anything specific — Veo has to guess what you mean. Precise terminology isn’t jargon for its own sake here, it’s the actual lever that produces a controlled result.

The core moves

Move	What it does	Best for
Static / locked-off	No camera movement at all	ASMR, product hero shots, dialogue close-ups
Dolly-in / dolly-out	Camera moves straight toward or away from the subject	Building tension, revealing scale, product reveals
Orbit	Camera circles around the subject, subject stays centered	Product 360°, hero object reveals
Pan	Camera pivots horizontally from a fixed point	Following action across a wide scene
Tilt	Camera pivots vertically from a fixed point	Revealing height, looking up at a building or object
Crane / boom	Camera moves vertically (and often forward) on an arm	Dramatic reveals, rising above a scene
Tracking shot	Camera moves alongside a moving subject, keeping pace	Following a walking or running subject
Push-in	Similar to dolly-in but often paired with a slow zoom	Emotional emphasis on a face or detail
FPV / handheld	Camera moves as if worn or carried by a person, some shake	Vlogs, street interviews, UGC, action sequences
Whip pan	Fast horizontal pan, often used as a transition	Cuts between scenes, energetic edits

Before / after: vague vs. named

Vague (unpredictable result):

The camera moves around the car in a cool cinematic way.

Named move (predictable result):

Slow orbit around the car, starting from the front-left and circling
clockwise to the rear-three-quarter angle. Shallow depth of field.

The second version tells Veo the direction, the starting point, and the endpoint — all things the vague version left for the model to guess, which is why results from the first version vary wildly between regenerations.

Stacking moves: why one is better than two

A common mistake is writing “the camera orbits while dolly-ing in” or “a crane shot that also pans.” Veo can attempt this, but combined moves are the single biggest cause of warped geometry and inconsistent subjects in Veo outputs — the model is trying to solve two simultaneous camera problems plus the subject and scene, and something gives. Pick the one move that matters most for the shot; if you need both a reveal and a push-in, that’s usually two beats (see the JSON prompt format guide for how to sequence multiple camera moves across timestamped scenes instead of one prompt).

Pairing camera moves with framing

A camera move alone is incomplete — pair it with a framing distance:

Medium close-up, slow push-in on the subject's face.
Wide shot, static, subject small in frame against a large landscape.
Extreme close-up, static, macro lens on a droplet forming.

“Push-in” without a framing distance leaves Veo to guess how far to push; specifying “medium close-up” gives it a starting and ending point.

Lens and depth of field control

Camera moves pair naturally with lens language:

Shallow depth of field — subject sharp, background soft. Good for product and portrait work.
Wide-angle lens — more of the scene in frame, slight distortion at edges. Good for environments and establishing shots.
Macro lens — extreme close focus. Good for texture, ASMR, product detail.
Deep focus — everything sharp front to back. Good for wide landscape or architectural shots.

Worked examples by move

Dolly-in tension shot:

A quiet cinematic hallway at night, one door slightly open at the far
end. Slow dolly-in from a medium-wide shot toward the door, low-key
blue lighting, shallow depth of field, dust visible in the light beam.
SFX: distant room tone, faint floorboard creak. 8 seconds.

This exact prompt is verified in the Veo Prompt Library.

Orbit product reveal:

Slow 180-degree orbit around a matte-black product on a seamless
grey studio backdrop, crisp edge highlights, shallow depth of field,
high-key studio lighting. 8 seconds, 16:9.

FPV handheld street shot:

Vertical 9:16 handheld camera at golden hour, following a host as they
approach a stranger on the street. Natural walking motion, slight
handheld sway, casual documentary feel. 8 seconds.

Common mistakes

Stacking moves (see above) — pick one.
No framing distance stated — “slow pan” alone is vaguer than “wide shot, slow pan left to right.”
Camera language that doesn’t exist in film vocabulary — “swooshy camera,” “dynamic movement” — replace with a named move.
Forgetting the move needs a direction — orbit needs clockwise/counterclockwise or a start angle; pan needs a direction.

FAQ

Does Veo 4 support more advanced camera moves than 3.1? Reporting suggests improved cinematography-aware camera control is one of the expected Veo 4 capabilities, but as of July 2026 Veo 4 has not shipped in a form we’ve been able to verify directly — treat any specific claim about Veo 4 camera syntax as unconfirmed until it ships. The moves in this guide are confirmed working on Veo 3.1.

Can I control exact camera speed (e.g., “2 seconds per orbit”)? Not precisely — Veo doesn’t take frame-accurate camera speed parameters. You can influence pace with words like “slow,” “quick,” or “snap,” and with JSON scene timing (see the JSON format guide) to bound how long a move has to happen.

What’s the difference between “dolly-in” and “push-in” if they sound similar? In practice with Veo, both terms produce a similar forward camera move — a physical dolly move vs. an optical zoom. If you want to be precise about which one, add “zoom” for an optical push or “camera physically moves forward” for a dolly, but either term reliably signals “get closer to the subject.”

Why does my orbit shot warp the subject’s shape? Orbits around complex or asymmetric subjects (people, animals) are harder for Veo to hold consistent than orbits around simple, symmetric objects (bottles, boxes). If a subject warps mid-orbit, try a shorter arc (90° instead of 360°) or a static/tracking shot instead.

Try it in the Builder

The Veo Prompt Builder has a dedicated camera-move field pulled from this same vocabulary, so you can pick a move and framing without memorizing the terms above.

Testing camera moves across models to see which handles orbits or crane shots best? PixVerse is worth comparing against Veo for motion-heavy shots. This is an affiliate link — we may earn a commission at no extra cost to you.