Camera moves, defined

A dolly is not a zoom. A pan is not a truck. Video models blur these together because most writing about them does too — and a blurred instruction produces a blurred shot. Here is what each move actually is, in the language a camera department uses.

The distinction almost everyone gets wrong

Camera moves split into two families, and mixing them up is the single most common reason a generated shot does not match what was asked for.

The camera travels. Its position in space changes. Dolly, truck, crane, orbit, push in, pull out. Parallax changes — near objects slide past far ones.
The camera pivots. Its position is fixed and only its aim changes. Pan, tilt, roll. There is no parallax at all; the world does not shift against itself.

"Zoom" belongs to neither: nothing moves, the lens changes focal length. A zoom flattens the image. A push in does not. Ask for a zoom when you meant a push in and you lose the sense of entering the space, which is usually the entire reason the move was wanted.

Moves that travel

Push in: The camera moves toward the subject along its own axis. Tightens the frame while keeping the perspective of the original distance changing naturally. The workhorse move for building attention on a face or an object.
Pull out: The reverse: the camera retreats, revealing context around a subject the audience has already been looking at. Endings, reveals, and the shot that says "and here is where they actually are".
Dolly (also: truck, crab): A lateral slide. The camera moves left or right while holding its distance from the subject. Produces strong parallax, which is what makes it read as movement rather than a cut. Purists reserve "dolly" for in-and-out and "truck" for sideways; in practice most crews say dolly for both.
Orbit (also: arc): The camera arcs around the subject, staying at a constant distance and staying aimed at it. The background rotates behind a subject that barely moves in frame. Reads as scrutiny or revelation.
Crane up / crane down: Vertical travel. Crane up rises and usually pulls back, opening a scene out. Crane down descends from a high wide view into the action. The classic establishing and departing move.
Reveal crane: Rises over something in the foreground — a wall, a ridge, a crowd — so the scene beyond it appears from behind the obstruction. The foreground object is the point: the move only works if something is in the way to begin with.
Follow behind / follow ahead: The camera rides with a moving subject, either trailing it or leading it while travelling backwards. The subject stays roughly fixed in frame and the world moves past.
Fly alongside: Travels level with the subject on a parallel course, holding it in profile. Vehicles, runners, anything with a direction of travel worth reading.
Handheld track: A follow with the stabilisation removed — small irregular deviations in position and aim. Reads as urgency or documentary. Different from camera shake added afterwards, which sits on top of the image rather than in the camera's path.
Dutch push: A push in with the horizon rolled off level. The roll is the message: something is wrong. Rarely holds up for a whole sequence.

Moves that pivot

Pan left / pan right: The camera stays put and turns horizontally. Because the position is fixed, nothing moves against anything else — this is what separates a pan from a dolly on screen, and what makes the two impossible to substitute for one another.
Tilt up / tilt down: The same, vertically: the camera lifts or drops its gaze from a fixed position. A tilt up a building is not a crane up a building, and the difference is obvious on screen.
Pan between: A pan with a defined start and end: aim begins on one subject and swings to a second. Dialogue, reactions, cause and effect in a single unbroken take.

Framings, which are not moves

These describe where the camera is, not what it does. A shot can hold one of these perfectly still, and often should.

Static: Locked off on a tripod. Nothing moves. Frequently the right answer, and the option people forget they have.
Wide establishing: High and far, taking in the whole space so the audience knows where they are before anything happens in it.
Close up: Long lens, tight on the subject, background compressed and thrown out of focus.
Two shot: Holds two subjects in frame together, so the relationship between them is the subject.
Over the shoulder: Past the near subject's shoulder onto the far one. Anchors a conversation in a space and tells you who is being looked at.
POV: The camera occupies the subject's eyeline, looking at what they are looking at.
Low hero push: Wide lens near knee height, pushing in. Makes the subject tower. The standard heroic introduction, and a cliché worth spending deliberately.

Why video models get this wrong

A text-to-video model learns camera language from captions, and captions are written by people describing what a shot felt like rather than how it was made. "Sweeping shot" could be a crane, an orbit, or a drone flying past. "The camera moves in" could be a push in or a zoom. The model has seen all of those labelled the same way, so it averages them.

Subjects do not suffer from this. "A red car" is unambiguous — the model has seen thousands of red cars, all labelled red car. Camera direction is where the ambiguity concentrates, which is why prompts fail on motion far more often than on content.

Two things help. Name the move precisely and separately from the subject, and say what happens to the framing across the shot rather than just naming a feeling — "starts wide on the doorway, ends tight on her hands" constrains the result in a way "dramatic camera work" cannot. Beyond that, the reliable fix is to stop describing the motion and show it instead.

Frequently asked

What is the difference between a dolly and a zoom?

A dolly moves the camera through space, so near and far objects shift against each other. A zoom changes the lens from a fixed position, which flattens depth and produces no parallax. They look completely different on screen despite both making a subject larger.

What is the difference between a pan and a truck?

A pan pivots the camera from a fixed position. A truck moves the whole camera sideways. A pan across a room shows you the room from one viewpoint; a truck across the same room shows you around the furniture in it.

Is a push in the same as a dolly in?

Yes, in normal usage. Both mean the camera physically travels toward the subject. "Push in" is more common in direction, "dolly in" more common on set.

What does a Dutch angle do?

It rolls the camera so the horizon is no longer level. It reads as unease or disorientation, and it is one of the few camera choices an audience notices consciously.

Which camera move should I use for an establishing shot?

A wide establishing framing, usually static or with a slow crane down or drift. The point is to let the audience read the space, and motion competes with that.

Blockshot takes these 25 named moves as instructions, builds the shot in 3D, and exports a reference clip a video model can follow. See how it works →