Big Idea

The argument in one line.

The gap between what you type and what AI video renders closes the moment you swap vague descriptions for the 42 cinematography terms that every major AI video model has been trained to recognize.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

You use AI video generators like Runway, Kling, or Sora and keep getting the wrong camera motion despite trying different prompts.
You know the shot you want visually but not the cinematography term that tells AI which move to execute.
You are building a reusable prompt library for video production and want a tested reference for motion vocabulary.
You direct AI-generated scenes and need shared terminology to communicate camera behavior to collaborators.

SKIP IF…

You want a guide to writing complete AI video prompts — this covers camera motion vocabulary only, not lighting, subject, or scene description.
You do not currently use any AI video generation tools and have no near-term plans to start.

TL;DR

The full version, fast.

Typing camera moves forward gives an AI video generator three valid interpretations: dolly, zoom, or hyperlapse — and which one fires is unpredictable. This video tests 42 named movements across 9 categories (dollies, zoom effects, tripod moves, slider moves, orbital shots, crane and pedestal, optical lens effects, drone aerials, and stylized moves), showing a real AI-generated clip for each. The result is a portable reference that replaces guesswork with the same vocabulary a film director uses to brief a camera operator — and it works because AI models are trained on content that uses those exact terms.

Members feature

Chat with this breakdown.

Modern Creator members can chat with any breakdown — ask for the hook, quote a framework, find the exact transcript moment. Unlocks at T2: refer 3 friends + add your own API key.

Create a free account →

Chapters

Where the time goes.

00:00 – 00:33

01 · Intro and Promise

Hook on the vocabulary problem; promise of 42 categorized movements with live AI clips

00:33 – 01:10

02 · Dolly Moves

Slow dolly in, slow dolly out, fast dolly in, vertigo effect — four foundational push/pull variations

01:10 – 01:50

03 · Zoom Effects

Infinite scale continuity, extreme macro zoom, cosmic hyperzoom — scale-compression moves

01:50 – 02:47

04 · Character-Mounted and Environmental Framing

Over the shoulder, fisheye/peephole lens, reveal from behind, wipe movement, fly-through aperture

02:47 – 03:14

05 · Focus and Lens Manipulation

Reveal from blur/fade in, rack focus foreground to background

03:14 – 03:42

06 · Tripod Moves

Tilt up, tilt down — the two fundamental rotational tripod axes

03:42 – 04:10

07 · Slider Moves

Camera truck left, lateral truck right — lateral translation on a track

04:10 – 04:48

08 · Orbital Movements

Orbit 180, fast 360 orbit, slow cinematic arc — circular camera paths around a subject

04:48 – 05:44

09 · Vertical Movements: Crane and Pedestal

Pedestal down, pedestal up, crane up high angle reveal, crane down landing

05:44 – 06:20

10 · Optical Lens Effects

Smooth optical zoom in, smooth optical zoom out, snap zoom

06:20 – 07:14

11 · Drone Aerial Views

Drone flyover, epic drone reveal, large scale drone orbit, top down god's eye view, FPV drone aggressive dive

07:14 – 09:48

12 · Stylized and Dynamic Movements

Handheld documentary, whip pan, Dutch angle, leading/following/side tracking, POV walk, hyperlapse, bullet time, barrel roll, worm's eye tracking

09:48 – 10:15

13 · Outro

Save the reference; comment which movement you use most

Atomic Insights

Lines worth screenshotting.

Typing camera moves forward gives AI video three valid interpretations: dolly, zoom, or hyperlapse — and the model picks one arbitrarily without a precise term.
A dolly physically moves the camera through space; a zoom changes focal length while the camera stays still — AI models treat these as completely separate instructions.
The vertigo effect (simultaneous dolly out and zoom in) is a named technique most people attempt to describe and consistently prompt incorrectly.
Speed adjectives are load-bearing in orbital prompts: slow cinematic arc and fast 360 orbit produce shots with completely different emotional registers.
Drone vocabulary has the widest spread of any category: flyover, epic reveal, god's eye view, and FPV dive all trigger visually distinct AI outputs despite all being aerial shots.
Combining two synonymous terms in a prompt — barrel roll, vortex inception shot — doubles the signal for moves that AI models interpret inconsistently.
Dutch angle is a single adjective that shifts an entire clip from neutral to tense without changing anything else in the prompt.
Rack focus is one of the highest-impact single-shot storytelling tools available in AI video and is fully promptable with a specific phrase.
Hyperlapse and time-lapse are not interchangeable: moving time lapse explicitly cues both camera travel and speed compression together.
Bullet time requires its own vocabulary — generic slow-motion prompts do not produce the frozen-moment multi-angle effect.
Extreme macro zoom and cosmic hyperzoom generate visually distinct results despite both being scale-push moves — the adjective carries the instruction.
POV walk and first-person walk are near-synonyms worth stacking in the same prompt when AI consistency on that move matters.

Takeaway

The vocabulary that makes AI video prompts actually work.

WHAT TO LEARN

Named cinematography terms are the exact instructions AI video models were trained on — using them instead of plain descriptions is the single biggest unlock for consistent output.

Vague descriptions like camera moves forward give AI video generators three equally valid interpretations — dolly, zoom, or hyperlapse — and the model picks arbitrarily unless you name the move.
The 9 movement categories in this video are mutually exclusive in how AI models interpret them; knowing which category you need narrows your prompt before you write a single word.
Speed adjectives are load-bearing for orbital and dolly moves — slow cinematic arc and fast 360 orbit trigger completely different emotional registers even though the move type is identical.
Drone vocabulary has the widest spread of any single category: flyover, epic reveal, god's eye view, and FPV dive each produce visually distinct outputs despite all being aerial shots.
Combining two synonymous terms in a single prompt — barrel roll, vortex inception shot — reinforces intent for moves that AI models interpret inconsistently across generations.
The vertigo effect (dolly zoom) is one of the most commonly mis-prompted moves; its proper name resolves the ambiguity that causes most failed attempts.
Time and speed manipulation terms are not interchangeable: hyperlapse, bullet time, and hyperzoom each trigger a specific motion algorithm and confusing them produces unrecognizable results.

Glossary

Terms worth knowing.

Dolly: Camera movement where the entire camera physically travels forward or backward through space on a track or wheeled platform. Produces natural parallax between foreground and background, distinct from a zoom which changes focal length without moving the camera body.
Vertigo Effect (Dolly Zoom): Simultaneous dolly out and zoom in (or vice versa) that keeps the subject the same apparent size while the background perspective shifts dramatically. Named for its use in Hitchcock's Vertigo; also called the Jaws effect.
Rack Focus: Shifting the lens focus point from one subject plane to another within a single continuous shot — typically foreground to background — to redirect the viewer's attention without a cut.
Truck Shot: Camera moves sideways on a physical track (left or right) while the lens stays pointed in the same direction. Distinct from a pan, which rotates the camera body on a fixed axis without lateral travel.
Pedestal: Vertical camera movement straight up or down while keeping the lens aimed horizontally. Distinguished from a crane shot by the absence of a boom arm arc — the camera travels in a pure vertical line.
Crane Shot: Camera raised or lowered on a boom arm, allowing sweeping vertical reveals or dramatic descents into a scene. The arc path differs from the straight-line travel of a pedestal move.
Orbit / Arc Shot: Camera circles around a fixed subject while continuously pointing at it, completing a partial or full rotation to create a 3D reveal of the subject and its environment.
FPV Drone: First-person view drone footage shot with a racing-style drone, typically fast and aggressive through tight spaces. Produces a physically impossible flying-through sensation distinct from stabilized aerial footage.
Dutch Angle: Camera tilted on its roll axis so the horizon appears diagonal in frame. Used to create psychological unease, tension, or disorientation without changing subject or lighting.
Hyperlapse: A time-lapse where the camera physically moves between frames — combining speed compression with spatial travel — as opposed to a static time-lapse where the camera stays fixed.
Bullet Time: Visual effect that freezes the action while the camera continues to move through space, simulating a multi-camera array capturing a single frozen moment from multiple angles. Made famous by The Matrix.
Snap Zoom: A rapid, hard optical zoom that punches in on a subject in a fraction of a second, creating percussive visual impact used in action sequences or comedic beat punctuation.
Worm's Eye View: Extreme low-angle shot positioned at or near ground level looking upward, which exaggerates the scale and perceived power of subjects above the camera.

Quotables

Lines you could clip.

00:13

“To get control over your generations, you need the right vocabulary.”

Punchy standalone thesis — no setup needed→ TikTok hook↗ Tweet quote

00:07

“You type camera moves forward, but do you get a dolly, a zoom, or a hyperlapse?”

Rhetorical hook that surfaces the pain immediately→ IG reel cold open↗ Tweet quote

The Script

Word for word.

00:00If you are generating AI video, you know that getting the motion right is the hardest part. You type camera moves forward, but do you get a dolly, a zoom, or a hyperlapse?

00:11To get control over your generations, you need the right vocabulary. Today, I'm testing and visualizing 42 specific camera movement prompts.

00:21We are going to look at everything from zollies to barrel rolls so you can see exactly how the AI interprets these commands. I've organized these by category to help you build your prompt library. Dolly moves.

00:35Slow dolly in.

00:48Slow dolly out.

01:01Fast dolly in.

01:11Vertigo effect.

01:24Infinite scale continuity, extreme macro zoom, cosmic hyperzoom, Character mounted framing.

01:51Over the shoulder.

01:57I still can't believe you said that to him. It had to be done. Besides, look how well it turned out.

02:05Fisheye or peephole lens.

02:12Hello. I have your delivery.

02:18Obstacle and environmental interaction. Reveal from behind. White movement.

02:34Thru shot, fly through aperture.

02:47Focus and lens manipulation. Reveal from blur. Fade in.

03:03Rack focus. Foreground to background.

03:14Tripod moves. Tilt up.

03:28Tilt down.

03:42Slider moves. Camera truck left.

03:57Lateral truck right.

04:10Orbital movements. Orbit 180.

04:22Fast 360 orbit.

04:35Slow cinematic arc.

04:48Vertical movements, crane or pedestal. Pedestal down.

05:05Pedestal up.

05:17Crane up. High angle reveal.

05:30Crane down, landing.

05:44Optical lens effects, smooth optical zoom in.

05:59Smooth optical zoom out.

06:12Snap zoom.

06:19Drone aerial views. Drone flyover.

06:35Epic drone reveal.

06:48Large scale drone orbit.

07:01Top down, god's eye view.

07:14FPV drone aggressive, drone dive.

07:27Stylized and dynamic movements. Handheld documentary style, whip pan.

07:48We have to accelerate the timeline on this project. Dutch angle.

07:57Clear the hallway.

08:01Emergency.

08:05Subject tracking. Leading shot. Backward tracking.

08:17Following shot, forward tracking.

08:24Side tracking parallel.

08:37POV walk, first person walk.

08:50Time and speed manipulation. Hyperlapse, moving time lapse.

09:06Bullet time, frozen moment.

09:17Extreme orientation and perspective. Barrel roll, vortex inception shot.

09:33Worm's eye tracking, ground level.

09:48Hopefully, this helps you take better control of your AI video output. If you found this list useful, save it for reference and let me know in the comments which movement you use the most. See you in the next one.

The Hook

The bait, then the rug-pull.

Every AI video prompt hides a vocabulary trap: type the wrong word for the camera move you want and the model gives you something technically plausible but visually wrong. This video solves that problem by testing 42 distinct camera movement terms against live AI generations so you can see, precisely, what each command produces.

Frameworks