AI Camera Control Glossary: Lens, Move, and Shot Terms for Previs (2026)

Camera language is the fastest way to get a specific shot out of an AI video model or a previs tool. A prompt that says "camera moves in" leaves speed, path and end position open, so the model fills them in differently on each run. This glossary defines 25 terms in three groups: lens terms, camera movement terms and shot framing terms. Each entry starts with a short, tool-neutral definition, followed by an example of how the term works in a 3D scene. The examples use Intangible's Compose mode, where cameras, lenses and moves are set in a real 3D scene. Film students, previs artists and creators new to camera work can use it as a quick reference.

What Is AI Camera Control?

AI camera control is any method of telling a video model where a virtual camera sits and how it moves during a shot. Methods range from text prompts and preset camera moves to cameras placed in a full 3D scene. Text alone leaves speed, path and end position open, and AI video models have no persistent understanding of scene state, so results vary from run to run. The more precisely the camera is specified, the more repeatable the shot.

Previsualization (previs) is the practice of planning camera angles, staging, lenses and timing before production, usually with simple 3D scenes or animated storyboards. For a short definition of the category, see AI storyboarding and previsualization.

Why Camera Terminology Matters for AI Video in 2026

Camera vocabulary has moved from film school to everyday creative work. A creator who knows what a 70mm lens does to a background, how a dolly differs from a zoom, or what a 2.39:1 frame signals can write clearer prompts and set up 3D cameras with intent. The terms also make feedback faster: "push in on a 50mm and end on a close-up" is easier to act on than "make it feel more intimate."

Text prompts and reference images do not tell AI video models how a scene changes over time, so precise terms help, and a camera placed in a 3D scene helps more. In tools that pass camera data to the model, the focal length, gate and move you choose become part of what the model receives rather than a description it has to interpret.

Lens Terms Every AI Video Creator Should Know

Focal Length

Focal length is the distance, in millimeters, between a lens's optical center and the sensor when focused at infinity. It sets the angle of view and magnification. Short focal lengths (14mm to 24mm) are wide and show more of the scene; standard lenses (35mm to 50mm) read close to natural vision; long lenses (70mm and up) magnify the subject and compress the background.

In Intangible: The Compose lens dropdown offers 14mm Ultra-Wide, 24mm Wide, 35mm Standard (the default), 50mm Normal, 70mm Portrait, 100mm Telephoto and 200mm Super Telephoto, plus Custom. Lens choice is saved per shot, and the focal length is named in the visualizer's auto-prompt.

Field of View (FOV)

Field of view is the angle of the scene a camera captures, set by focal length and sensor size. On a full-frame sensor, a 35mm lens covers about 54 degrees horizontally and a 70mm lens about 29 degrees. A wider field of view shows more environment; a narrower one isolates the subject. FOV can be measured horizontally, vertically or diagonally, so check which one a tool uses.

In Intangible: The Custom lens panel lets you switch the slider between focal length in millimeters and vertical field of view in degrees. Both views track the same value.

Film Gate

The film gate is the frame area a camera records, which sets the shape of the image. On film cameras it was a physical aperture in front of the film; in 3D tools it defines the camera's frame, and some tools set a separate resolution gate for output pixels. Choosing the gate before framing means compositions are built for the final format rather than cropped later.

In Intangible: The aspect ratio dropdown in Compose works as the gate. Changing it reframes the shot as a crop within the same camera setup, the auto-prompt names the gate, and downloads keep the chosen aspect ratio.

Aspect Ratio

Aspect ratio is the width-to-height relationship of the frame. It shapes how a shot reads before any lens choice: 9:16 feels phone-native, 1:1 feels editorial, 16:9 is the broadcast and streaming default, and 2.39:1 reads as theatrical widescreen. Pick the ratio for where the work will be seen, since recomposing for a different ratio later often changes what fits in frame.

In Intangible: Options are Vertical 9:16, Square 1:1, 4:3, European Widescreen 1.66:1, 16:9, Cinemascope 2.39:1 and Ultra Panavision 70 (2.76:1). Some video models support only certain ratios, and the visualizer warns you at generation time.

Depth of Field

Depth of field is the distance range in front of and behind the subject that appears acceptably sharp. It is controlled by aperture, focal length and camera-to-subject distance. A wider aperture, a longer lens and a closer subject all make depth of field shallower, which softens the background and isolates the subject. AI video models reproduce depth of field as a learned look, so describing it in the prompt helps.

In Intangible: Longer lenses such as 70mm and 100mm lean toward subject isolation and shallow-depth looks; pair the lens choice with a prompt that describes the background softness you want.

Anamorphic Lens

An anamorphic lens squeezes a wide image horizontally onto the sensor or film, usually by a factor of 2, and the image is stretched back out in post. The result is a wide frame, commonly 2.39:1, with signature traits: oval out-of-focus highlights, horizontal lens flares and a distinctive edge falloff. These traits are strongly associated with theatrical feature films.

In Intangible: Anamorphic squeeze and flares are not simulated per lens. Choose Cinemascope 2.39:1 as the gate and request the anamorphic look in the prompt or through a style preset.

Clipping Planes

Clipping planes are the near and far distances beyond which a 3D camera stops drawing geometry. The near plane hides objects too close to the lens; the far plane hides objects too far away. Adjusting them fixes common previs problems, such as a camera placed inside a wall or distant clutter showing up behind a long-lens shot.

In Intangible: The Custom lens panel sets near and far clipping (defaults 0.1 and 10,000) and can draw both planes in the viewport.

Camera Movement Terms for AI Video and Previs

Dolly (Push-In and Pull-Back)

A dolly move travels the whole camera toward or away from the subject. Moving in is a push-in or dolly-in; moving out is a pull-back or dolly-out. Because the camera itself moves, foreground and background shift at different rates, creating parallax and a sense of depth. That parallax is what separates a dolly from a zoom. Push-ins often build tension or draw attention.

In Intangible: Keyframe the camera's position at a far point and a near point on the timeline, or use the dolly handle in fine-tune mode for small, precise moves.

Zoom

A zoom changes the lens's focal length while the camera stays in place, so the image enlarges or shrinks without the camera moving. There is no parallax: subject and background scale together, and the image flattens as focal length increases. Zooms feel more mechanical than dollies and are often used for quick emphasis, documentary energy or a deliberately retro look.

In Intangible: Lens is a keyframable camera property, so a zoom can be animated separately from camera position.

Pan

A pan rotates the camera horizontally, left or right, from a fixed position. The camera does not travel, so there is no change in perspective between near and far objects. Pans follow a subject moving across the frame, scan a location or reveal information at the edge of the frame. A fast pan between subjects is called a whip pan.

In Intangible: Pan with rotation keyframes on a free camera, or let an aim target handle rotation automatically.

Tilt

A tilt rotates the camera vertically, up or down, from a fixed position. Like a pan, it changes where the camera points without moving the camera. Tilting up a building emphasizes height; tilting down from a face to an object shifts attention. A tilt is often confused with a boom, which raises or lowers the whole camera instead of angling it.

In Intangible: Tilt with rotation keyframes on a free camera. With an aim target, rotation follows the target, so leave tilt to the target lock.

Boom (Pedestal)

A boom, also called a pedestal move, raises or lowers the entire camera through space while it keeps pointing in the same general direction. Unlike a tilt, the camera's height actually changes, so the perspective on the subject shifts. Rising booms can reveal scale or a wider environment; descending booms can settle into a scene or a detail.

In Intangible: Keyframe the camera's vertical position, or use the pedestal handle in fine-tune mode. Booms work with an aim target active.

Truck

A truck moves the camera sideways, parallel to the subject, rather than toward or away from it. On set, the camera platform rolls left or right on a track or dolly. The background slides past while the subject stays roughly in place, which suits walking-and-talking scenes and slides across product displays. A truck is different from a pan, which only rotates the camera.

In Intangible: Keyframe lateral camera position or use the truck handle in fine-tune mode. With an aim target, the camera trucks while staying pointed at the subject.

Tracking Shot

A tracking shot moves the camera alongside, ahead of or behind a moving subject, holding the subject in roughly the same place in frame while the environment passes by. It puts the viewer in the subject's motion. On set, tracking shots use dollies, Steadicams, gimbals or vehicles. The term is sometimes used loosely for any camera move that follows action.

In Intangible: Bind the camera to an aim target on the moving subject and keyframe the camera's position along its route, or parent the camera to a motion path.

Crane Shot

A crane shot mounts the camera on a crane, jib or technocrane so it can rise, fall and sweep through space in one continuous move. It can start low and rise to reveal a whole location, or descend from overhead into an intimate frame. Crane moves combine boom, truck and often pan in a single gesture, which makes them feel grand and deliberate.

In Intangible: Combine vertical and lateral position keyframes, or parent the camera to a motion path and keyframe its Progress value. A game controller's shoulder buttons also control elevation for long crane-style moves.

Arc Shot (Orbit)

An arc shot moves the camera along a curved path around the subject while keeping it in frame, so the background rotates behind it. A full 360-degree circle is called an orbit. Arcs show a subject from several sides in one continuous move, which makes them popular for product reveals, hero moments and character introductions.

In Intangible: Parent the camera to a circular motion path and keyframe its Progress, with an aim target keeping the subject centered. You can also ask AI Composer to orbit the camera, and the resulting keyframes stay editable.

Dutch Angle (Dutch Tilt)

A Dutch angle rolls the camera on its lens axis so the horizon sits at a diagonal. The tilted frame signals unease, disorientation or instability, and it is common in thrillers, horror and stylized action. A roll can be held for the whole shot or animated so the frame tips over or rights itself.

In Intangible: Roll is keyframable, including on cameras bound to an aim target, and fine-tune mode has a dedicated roll handle.

Dolly Zoom (Vertigo Effect)

A dolly zoom moves the camera toward or away from the subject while zooming the lens in the opposite direction. The subject stays about the same size in frame while the background appears to stretch or compress, creating an unsettling shift in perspective. It is often called the vertigo effect after its use in Alfred Hitchcock's 1958 film Vertigo.

In Intangible: Camera position and lens are both keyframable on the same shot, so you can animate the two together and preview the effect before rendering.

Shot Framing Terms for AI Video and Previs

Establishing Shot

An establishing shot opens a scene by showing where the action takes place. It is usually a wide or extreme wide shot that reveals the location, architecture or landscape before the camera moves closer. It orients the audience and sets mood, time of day and scale, so later, tighter shots make spatial sense.

In Intangible: Start a sequence with a camera on a 24mm Wide lens, which Intangible's guidance recommends for establishing shots where surrounding context matters.

Wide Shot

A wide shot shows the subject in full, head to toe, with a good amount of surrounding environment. The subject is readable, but the setting carries nearly equal weight. Wide shots suit action, geography, blocking and group scenes, and they help viewers understand where people and objects sit in relation to each other.

In Intangible: Each saved shot keeps its own camera position and lens, so a wide shot on 24mm or 35mm stays consistent across renders.

Medium Shot

A medium shot frames a subject from about the waist up. It shows enough body language to read gesture and emotion while keeping some background in view, which makes it the workhorse of dialogue, interviews and presentations. It balances character and context without the intensity of a close-up.

In Intangible: Save a 35mm or 50mm camera at waist height as its own shot so the framing repeats exactly when you return to it.

Close-Up

A close-up fills the frame with a face, a hand, a product or another detail. It removes most of the environment and focuses attention on expression, texture or significance. Longer lenses are common for close-ups because they compress the background and isolate the subject. An extreme close-up goes tighter still, often on eyes, a label or a mechanism.

In Intangible: The 70mm Portrait lens is the default suggestion for product heroes and beauty framings, with 100mm for stronger isolation.

Over-the-Shoulder Shot (OTS)

An over-the-shoulder shot frames one subject from behind another, with part of the foreground person's head or shoulder in frame. It places the viewer inside a conversation and establishes eyelines and the spatial relationship between two characters. OTS shots are usually cut in matching pairs, one from each side.

In Intangible: Place a camera behind one character, aim it at the other with an aim target, and save a matching reverse as a second shot in the same scene.

Point-of-View Shot (POV)

A point-of-view shot places the camera where a character's eyes would be, showing what that character sees. It creates instant identification with the character and is common for discoveries, threats and first-person moments. Camera height, lens and movement all shape how convincing the POV feels.

In Intangible: Position the camera at the character's eye height. Game Mode also lets you walk a scene at eye level to scout POV angles.

Two-Shot

A two-shot frames two subjects together, usually at medium distance, so both are readable. The relationship between them is the subject: spacing, overlap and which one is closer to the lens all carry meaning. Two-shots are common in dialogue and negotiation scenes and are often intercut with singles and over-the-shoulder shots.

In Intangible: Build the two-shot, OTS pair and close-ups as separate shots on the same 3D scene, so every angle shares one layout.

How Intangible Puts These Terms into Practice

This section shows how the terms above map to controls in Intangible, a creative platform for development teams in retail, marketing and deep tech, along with film, advertising and architecture teams. Its structured guided workflows require no advanced expertise.

Placing and Configuring the Camera

In Compose, free cam lets you move quickly through the scene to find an angle, and fine-tune mode gives small, precise moves with rosette handles for pan, dolly, truck, pedestal and roll. Xbox and PlayStation controllers are supported for long, crane-style moves. Each shot saves its own camera, lens and aspect ratio, so one scene can hold a 24mm establishing shot, a 50mm medium and a 100mm close-up.

Animating Camera Moves

Camera position, rotation, lens and aim target are keyframable, along with object position, rotation and scale. Interpolation modes include easing, smooth, linear and step. An aim target keeps the camera pointed at a subject while you keyframe position and roll, and any camera can be parented to a motion path whose Progress value is keyframable. AI Composer can also create moves from a prompt, such as orbiting the camera, and the keyframes it writes stay editable.

Rendering Across AI Video Models

The finished shot can be rendered with Luma, Kling, Seedance, Wan, Omni or Google's Veo. Switching does not require re-prompting, because Intangible translates prompts for each model in the background. AI video models have no persistent understanding of scene state, so the 3D scene supplies it: the auto-prompt names the lens and gate, and the model receives the camera framing as well as the text. Drift is reduced, not removed, and results are closest to plan when the 3D scene, the image reference and the prompt agree.

Key Takeaways

• Lens terms (focal length, field of view, gate, aspect ratio, depth of field) decide how a frame looks before anything moves.
• Movement terms (dolly, zoom, pan, tilt, boom, truck, tracking, crane, arc, Dutch angle, dolly zoom) describe how the camera changes over a shot. A dolly moves the camera; a zoom changes the lens.
• Framing terms (establishing, wide, medium, close-up, OTS, POV, two-shot) describe what the frame contains.
• Text prompts and reference images do not tell AI video models how a scene changes over time, so precise camera terms help, and cameras placed in a 3D scene help most. Teams that use Blender or another 3D tool to feed models like Seedance pass only a flat 2D frame of the scene, while Intangible passes the full 3D data of the scene to the video model.

To try these terms on a real 3D camera, open Compose in Intangible, or see how it compares in the best previs software for 2026.

FAQs About Camera Terms for AI Video and Previs

What Camera Terms Do You Need to Control AI Video?

Start with focal length, aspect ratio, the core moves (dolly, zoom, pan, tilt, truck, boom and orbit) and the main shot sizes (wide, medium and close-up). Those cover most shot descriptions. Add depth of field, Dutch angle and dolly zoom when you need a specific look or feeling.

What Is a Push-In Shot?

A push-in moves the camera closer to the subject during the shot. Because the camera travels rather than the lens zooming, the background shifts relative to the subject, creating parallax and depth. Push-ins draw attention and build tension. In a 3D tool such as Intangible, you set one by keyframing the camera at a far position and then a near position.

Which Lens Focal Length Should I Use for Previs?

35mm is a reliable default because it reads naturally and does not impose a strong look. Use 24mm for establishing shots that need context, 50mm when a shot should feel observed, 70mm for product heroes and portraits, and 100mm or 200mm for strong compression or distant action.

What Is a Film Gate in 3D?

In 3D software, the film gate is the camera's frame area, which sets the shape of the image. Some tools also have a separate resolution gate for output pixels. In Intangible, the aspect ratio dropdown in Compose acts as the gate: changing it reframes the shot within the same camera setup, and renders arrive at that ratio.

What Is the Difference Between a Dolly and a Zoom?

A dolly moves the camera through space, so foreground and background shift relative to each other. A zoom changes focal length from a fixed position, so everything scales together with no parallax. Dollies feel immersive; zooms feel flatter and more mechanical. Combining the two in opposite directions creates the dolly zoom.

What Is the Difference Between a Pan, a Tilt, a Boom and a Truck?

A pan rotates the camera left or right, and a tilt rotates it up or down, both from a fixed position. A boom raises or lowers the whole camera, and a truck moves it sideways. Pans and tilts change where the camera points; booms and trucks change where the camera is.

How Do I Keep a Camera Locked on a Moving Subject?

Use an aim target or look-at constraint, which rotates the camera to follow the subject automatically. In Intangible, you bind the shot's camera to an aim target on the subject in Compose, then keyframe the camera's position and roll as needed. Avoid keyframing rotation directly, since it fights the target lock.

What Aspect Ratios Are Used for Film and Social Media?

Vertical 9:16 is standard for Reels, TikTok and Shorts, and 1:1 for some feeds. 16:9 is the broadcast and streaming default. Theatrical films commonly use 1.85:1 or 2.39:1, and 2.76:1 is the Ultra Panavision 70 format. Intangible supports 9:16, 1:1, 4:3, 1.66:1, 16:9, 2.39:1 and 2.76:1.

What Is a Dolly Zoom, and Can I Create One in a 3D Previs Tool?

A dolly zoom moves the camera in one direction while zooming the lens in the other, keeping the subject the same size as the background stretches or compresses. In any 3D tool where both camera position and focal length are keyframable, including Intangible, you can animate the two together and preview the effect.

What Is Depth of Field, and How Does It Affect AI Video?

Depth of field is the range of distance that appears sharp. Longer lenses, wider apertures and closer subjects make it shallower. AI video models produce depth of field as a learned look, so describe it in the prompt alongside a longer focal length for a soft background.

What Shot Types Work Best for Product Visualization?

A wide or establishing shot places the product in context, a close-up on a 70mm or 100mm lens shows detail and finish, and an arc or orbit shows the product from every side in one move. Building all three on the same 3D scene keeps the product and set consistent across angles.

Can I Animate an Orbit Shot Around a Product?

Yes. In most 3D tools you move the camera along a circular path while it stays aimed at the product. In Intangible, parent the camera to a circular motion path, keyframe its Progress and bind it to an aim target on the product, or ask AI Composer to orbit the camera and adjust the keyframes it creates.