AI video prompt broken into subject, scene, motion, camera, timing, and sound planning cards

How to Write AI Video Prompts That Produce Clearer Motion

August 16, 2026

A useful AI video prompt answers two questions: what should the viewer see, and how should it change over time? When those instructions are mixed into an abstract paragraph, the result may look polished while missing the intended action.

Use this structure for one shot at a time. For the complete planning process, begin with the AI video prompts guide.

1. Name the Type of Shot

Start with the intended output: cinematic live action, stop motion, product demonstration, illustrated animation, documentary-style footage, atmospheric b-roll, or another clear category.

This establishes expectations for texture and motion. Avoid combining several incompatible styles unless the transformation is the purpose of the shot.

2. Describe the Subject Precisely

Identify the main subject with details that matter on screen: appearance, clothing, material, shape, scale, condition, or relationship to other objects.

Do not spend most of the prompt on invisible backstory. “A tired courier in a rain-darkened yellow jacket” gives the model more visual direction than a paragraph about the courier’s childhood.

3. Establish the Environment

Describe the location, time, weather, important surfaces, background activity, and depth. Give the subject somewhere to move.

Choose details that affect the shot: wet pavement for reflections, a narrow corridor for constrained motion, or a windy field for visible environmental response.

4. Use Concrete Action Verbs

Describe what happens physically. Replace “shows determination” with “tightens her grip, looks toward the ridge, and takes one step forward.” Replace “the product feels luxurious” with visible materials, controlled movement, and lighting.

Keep one dominant action. If the prompt contains five events and a transformation, split it into separate shots.

5. Add Environmental Motion

State what moves besides the subject: curtains lift in a draft, steam curls upward, dust trails behind a vehicle, reflected lights pass across glass, or people continue walking in the distance.

One or two supporting motions can make the shot feel alive without overwhelming it.

6. Choose Framing and Camera Motion

Specify a wide, medium, close, overhead, low-angle, or point-of-view shot when framing matters. Then choose a camera behavior such as locked, pan, tilt, dolly, orbit, crane, handheld tracking, or slow push-in.

Use the camera-movement guide for AI video to match the move to the purpose. Avoid asking for a locked camera and a sweeping orbit in the same shot.

7. Control Pace and Timing

Use words that describe speed and rhythm: slowly, sharply, with a brief pause, continuous movement, gentle acceleration, or held final frame.

When order matters, write a short beginning, change, and ending. Do not script more action than the clip duration can show clearly.

8. Define Lighting, Color, and Style

Describe the light source and quality: soft window light, hard noon sun, warm practical lamps, neon reflected on wet pavement, or a narrow spotlight through haze.

Add a restrained palette and medium. Visual style should support the subject and motion rather than become a pile of fashionable adjectives.

9. Add Audio Separately When Supported

If the current tool supports audio, include a separate line for ambience, effects, dialogue, or music. Keep spoken lines short. Confirm the tool’s current capabilities because audio support and controls change.

A Reusable AI Video Prompt Structure

Shot type and style: [type]
Subject: [visible description]
Environment: [place, time, important background]
Subject action: [one clear action]
Environmental motion: [one or two supporting movements]
Camera: [framing plus one movement]
Timing: [pace, sequence, final hold]
Lighting and color: [direction]
Audio, if supported: [ambience, effects, short dialogue]
Continuity requirements: [approved appearance, props, or references]

Example: From Vague to Directable

Vague:

Make a beautiful cinematic video of a coffee shop that feels inspiring.

More directable:

Cinematic live-action medium-wide shot inside a quiet neighborhood coffee shop at sunrise. A barista places a ceramic cup beneath the espresso machine and turns the handle once. Steam rises into warm side light while two customers move softly out of focus in the background. The camera makes a slow, steady push toward the cup and holds as the first drop falls. Amber, cream, and dark green palette. Natural room ambience and a soft machine hiss, if audio is supported.

The second version describes visible decisions without prescribing an entire film.

Common Problems and Fixes

Too many actions: divide the prompt into shots.
Random motion: use concrete verbs and direction.
Camera ignores the subject: simplify to one camera move.
Scene feels frozen: add restrained environmental motion.
Style changes: use an approved reference and continuity sheet where supported.
Timing feels rushed: remove events or increase duration if available.
Prompt revisions feel random: change one variable and record the result.

Official Runway guidance specifically recommends beginning with essential motion and adding detail during iteration. Google DeepMind’s Veo guide likewise highlights framing, motion, style, lighting, character, location, action, and audio as controllable elements.

Writing and testing these details takes repeated generations. Browse the AI video prompt collection for prepared scene foundations, then adjust them for your tool and intended shot.

Frequently Asked Questions

Should an AI video prompt use keywords or sentences?

Clear natural-language sentences are a reliable default. Some interfaces support additional controls, but the visible and motion decisions still need to be coherent.