Camera Movement Vocabulary for AI Video (30 Terms)

TL;DR: Camera movement vocabulary (dolly, pan, tilt, crane, orbit, dolly zoom) gives a text-to-video model a specific instruction instead of a vague one. Below are 30 real terms, grouped by what they physically do, each with a correct definition, what it does to an AI shot, and a copy-paste fragment, plus where AI-video prompting has drifted from what the terms actually mean.

One honesty note first: Prompt Architects writes the prompt, not the video. Nothing here renders a clip; these are words to paste into whichever model does that.

What Is Camera Movement Vocabulary, and Why Does It Matter for AI Video?

Camera movement vocabulary is the language filmmakers use for how a camera changes position, angle, or focal length during a shot: dolly, pan, tilt, crane, zoom, orbit, and more. Each term names a specific piece of equipment or optical change, so it produces a learnable visual signature. Text-to-video models trained on footage captioned with this language, so naming the real technique beats a mood word like "cinematic."

The trouble: a lot of AI-video guidance, including some vendors' own, uses these words loosely: "zoom" for what's actually a dolly, "crane" for anything that rises, "tracking" for any sideways move. That's fine until you need a specific effect, and the vague version won't give it to you. Below: all 30 terms defined correctly, every drift from film-set meaning flagged, with a fragment for each.

For 30 full shot recipes that combine these moves with framing and lens choices for Veo 3 and Kling, see 30 Cinematic Camera Prompts for Veo 3 and Kling AI. For how camera movement fits alongside subject, action, setting, light, and sound, see the seven-part anatomy of a video prompt.

How Do You Combine Camera-Movement Terms in One Prompt?

Layer the subject and action first, then one dominant move, then framing and lens as support.

Which Terms Move the Camera Toward or Away From the Subject?

The film-set names first, then the words Kling's own guidance actually recommends.

Dolly In

Camera travels forward on a track or cart, closing the distance. Background objects shift past each other at different rates as it moves (parallax); perspective stays natural. slow dolly in on tracks toward the subject's face, background shelving parallax-shifting past the frame edges

Dolly Out

The reverse: the camera physically retreats, usually to reveal more space. camera dollies out from a close-up on the subject's hands to a wide shot of the workshop

Tracking Shot (Truck)

The camera moves laterally, parallel to the subject, rather than toward it; filmmakers also call the sideways version a truck. smooth tracking shot moving right alongside the subject as they walk, camera at shoulder height, matching their pace

Push-In

The everyday name for dolly in. Kling's prompting guide recommends "push in"; "dolly in" isn't on that list at all (kling.ai/blog/kling-ai-camera-control-video-guide, accessed Aug 27, 2026). slow push-in toward a subject seated at a desk, soft daylight, stable camera

Pull-Back

The everyday name for dolly out. The same Kling guide recommends it for reveals: "the first frame starts close and the final frame shows a larger setting." camera pulls back from a close-up on a ring to reveal the full jewelry display

Why Isn't Zoom the Same as Dolly?

The confusion that gives away a prompt written without knowing the craft. A dolly and a zoom both make the subject bigger; that's the only thing they share.

Zoom In

The camera doesn't move; the lens's focal length increases, magnifying the image. No parallax, since the camera stays put, and a longer focal length flattens depth. zoom in slowly on the subject's face, no camera movement, background compresses and flattens

Zoom Out

The reverse optical change: focal length decreases, the frame widens, depth stretches back out. zoom out from a tight close-up to reveal the full room, static camera position

Crash Zoom

A zoom executed in a fraction of a second rather than smoothly: a jolt, not a glide, for comedic or shock emphasis. crash zoom to an extreme close-up on the subject's eyes over 0.3 seconds, sudden and jarring

Kling's own broader guide actually keeps the two straight on close reading: "Dolly moves: physically move closer to or farther away from subjects" versus "Zoom actions: move viewers in or move them out" (kling.ai/blog/ai-camera-control-movement-prompts-guide, accessed Aug 27, 2026), physical movement versus a perceptual change. Worth noting since a model has no physical lens either way; both labels reproduce a signature learned from footage, not an actual rig or lens.

The clearest proof they differ: the dolly zoom below moves one while reversing the other. Identical moves would cancel to nothing.

Which Terms Sweep the Camera Sideways?

Pan

Camera stays fixed and rotates horizontally, left or right, like turning your head without moving your feet. slow pan left to right across the skyline, camera fixed on its axis, no forward movement

Whip Pan

An extremely fast pan, fast enough to blur into a streak for a few frames. Used mid-shot to snap attention from one subject to another within the same take. whip pan 120 degrees in 0.3 seconds from the speaker to the doorway, motion blur through the turn

Whip Pan Transition

The same move, used editorially: the blur masks a cut to a new location or time, disguising the edit as one continuous motion. match the blur of an exiting whip pan to an incoming whip pan revealing a new location, disguising the cut

Which Terms Move the Camera Up or Down?

Tilt pivots in place. The other three physically relocate the camera, and the rig used is what tells them apart.

Tilt Up

Camera stays fixed and rotates vertically, pointing higher: the vertical equivalent of a pan. camera tilts up from the subject's shoes to their face, fixed position, no lateral movement

Tilt Down

The reverse, rotating lower without moving the base. tilts down from a ceiling light fixture to the table below, static base

Pedestal

Camera rises or falls on a vertical column while the tilt angle stays the same, common in studio and broadcast work. pedestal up two feet, camera rises vertically with framing angle unchanged, subject stays centered

Crane Shot

Camera rides a long counterweighted arm, often motorized, capable of large sweeping moves at once: the move behind most big reveal shots. Kling's guide names it: "Crane movements rise up and over barriers" (kling.ai/blog/ai-camera-control-movement-prompts-guide, accessed Aug 27, 2026). crane up from ground level, sweeping over the crowd to reveal the full stadium

Jib Shot

A smaller, lighter counterbalanced arm doing the same job over a shorter range: a few feet rather than a full crane's reach. jib rises from a tabletop product shot up and over to reveal the storefront behind it

Which Terms Circle the Subject?

Orbit

Camera moves in a circular path at a constant distance, reframing to keep the subject centered, usually implying it keeps going rather than stopping. Kling's own list includes "orbit slowly" (kling.ai/blog/kling-ai-camera-control-video-guide, accessed Aug 27, 2026). slow 360-degree orbit around the subject, constant radius, subject centered throughout

Arc Shot

A partial orbit: a shorter sweep, often 90 to 180 degrees, revealing a different side of the subject rather than completing a circle. The two overlap heavily and Kling's list doesn't distinguish them, so state the degrees if you want one specifically. 90-degree arc from the subject's front to their profile, camera maintains a fixed radius

Which Terms Describe How Steady or Shaky the Camera Is?

Handheld

Camera held by hand or body, not stabilized: natural, slightly unsteady human movement, the documentary look. Kling's guide lists "handheld documentary style" among its style descriptors (kling.ai/blog/ai-camera-control-movement-prompts-guide, accessed Aug 27, 2026). handheld camera following the subject, natural sway and slight shake, documentary feel

Steadicam / Gimbal

A stabilizing rig, worn or carried, removing handheld shake while keeping organic movement: historically a spring-armed Steadicam harness, more often today a motorized handheld gimbal. Kling names "steady Steadicam movement" (same source). steadicam-smooth camera gliding alongside the subject through the hallway, no shake, floating movement

Locked-off

Camera fixed on a tripod, no movement at all. Kling's own vocabulary calls this "static camera." Say it explicitly, or models tend to add their own drift. locked-off static shot, camera fixed on a tripod, zero movement for the full duration

Which Terms Put the Viewer Inside the Scene?

POV Shot

The camera stands in for a character's own eyes, showing exactly what they'd see. first-person POV shot walking down the corridor, camera at eye height, hands occasionally entering frame

Over-the-Shoulder

Camera looks past one subject's shoulder at another, implying conversation or confrontation between them. over-the-shoulder shot from behind the first speaker, focus on the second subject's face, both partially in frame

Tracking-Follow

Camera moves with a subject from behind or beside, matching their speed and path. Distinct from Kling's Motion Control, which transfers a reference video's movement onto a character rather than moving the camera; see our full breakdown of Kling's camera and motion systems. tracking-follow shot from behind the subject, matching their walking pace, gimbal-smooth

Which Terms Tilt or Refocus the Frame, Rather Than Move It?

Both get filed under camera-movement vocabulary even though, strictly, neither is a movement.

Dutch Angle

Camera tilted on its roll axis so the horizon sits diagonally: a framing choice for unease or disorientation, often held on an otherwise static shot rather than being a movement itself. 15-degree Dutch angle, medium close-up, subject off-center leaning into the tilt

Rack Focus

Not a camera movement at all: a focus pull, historically done by a camera assistant turning the focus ring mid-shot. Sharp focus shifts from one subject or depth plane to another while the frame stays put. Grouped here because AI-video prompts phrase it the same way as an actual move: an instruction to "the camera." rack focus from the foreground object to the subject's face in the background, shallow depth of field

What Is the Dolly Zoom, and Why Does It Prove Zoom Isn't Dolly?

Dolly Zoom (Vertigo Effect)

Camera dollies one direction while the lens zooms the opposite direction at a matched rate, so the subject's size stays constant while the background stretches or compresses around them. Nicknamed the Vertigo effect after Alfred Hitchcock's 1958 film. It only works because dolly and zoom are genuinely different operations: moving closer while zooming out cancels the subject's size change but not the background's perspective shift. dolly forward while zooming out to a wider focal length, subject's size held constant, background stretching and warping around them

Which Terms Describe Aerial and Drone Camera Moves?

Aerial Establishing Shot

A high, usually static or slow-moving wide shot from above, opening a sequence before narrowing in. aerial establishing shot, high wide angle over the coastline, slow forward drift, no subject yet in frame

Drone Ascent

Camera rises vertically from near ground level to altitude: the aerial equivalent of a pedestal or crane up, untethered. Kling's guide lists "drone-like aerial viewpoint" (kling.ai/blog/ai-camera-control-movement-prompts-guide, accessed Aug 27, 2026). drone ascent from 20 feet to 300 feet above the subject, smooth vertical rise, landscape expanding into view

Drone Descent

The reverse: starting high, dropping toward a subject or ground point, often ending a sequence on a detail. drone descent from a wide aerial view down to eye level with the subject, smooth deceleration on arrival

Where Has AI-Video Camera Vocabulary Drifted From the Film Set?

Four patterns stand out: vendor guidance preferring push/pull over dolly, then using dolly anyway; "zoom" as a loose synonym for "closer"; pedestal, crane, and jib collapsing into "goes up"; and rack focus and Dutch angle requested as "movements" that never move the camera.

None of this makes the drifted phrasing wrong; it works well enough to be Kling's own default advice. It just means naming the film-set term precisely still matters when you need the exact effect.

What Do Full Camera-Movement Prompts Look Like Combined?

Each stays close to the one-dominant-move rule from 30 Cinematic Camera Prompts for Veo 3 and Kling AI: one movement doing the real work, framing and lens as support.

Wide aerial establishing shot, high angle over a coastal town at golden hour,
slow drone descent from 300 feet to 50 feet, warm low-angle light, long shadows.

Medium close-up, slow push-in toward the subject's face, 35mm lens, shallow
depth of field, soft window light from camera left, locked-off framing before
the push begins.

Medium shot, dolly forward while zooming out to a wider focal length (dolly
zoom), subject's size held constant, background warping and stretching behind
them, muted color palette.

Static medium shot for one second, then a fast whip pan 150 degrees in 0.3
seconds to a second subject entering frame, motion blur through the turn.

Tracking-follow shot from behind the subject walking through a market, gimbal-
smooth, 24mm lens, camera at shoulder height, background parallax passing at
natural walking speed.

What Should You Do Next?

Pick the three or four moves your project actually needs (a push-in for emphasis, an orbit for a showcase, a locked-off frame for anything deliberate) and write them with the correct term rather than the vague one. The vocabulary transfers across every model trained on real film footage; what varies is which words a given vendor recommends, and Kling is currently the only one whose list is readable on its own site. Movement is one axis; light is the other, and Lighting Vocabulary for AI Image Prompts does the same term-by-term job for it.