AI Video Prompt Guide
Learn how to write effective prompts for AI video generation. Structure, references, camera, motion, style and advanced examples — all in one place.
1. Prompt Structure
A strong prompt has three parts: what you’re referencing, what the video is about, and what happens on screen.
@Image 1 provides the character's appearance. A young woman walks through a quiet city street at sunset, cinematic style, slow dolly-in. 0-3s: Wide shot, she enters from the left, warm sunlight. 3-8s: Close-up, she looks up at the sky, camera slowly pushes in. 8-12s: She stops, wind moves her hair, soft ambient sound.
2. References
When you upload images, video or audio, describe clearly what each file is used for. This is the single biggest factor in getting consistent results.
| Reference | Purpose | How to write it |
|---|---|---|
| @Image 1 | Character appearance | Character reference (locks face / outfit) |
| @Image 2 | Scene / background | Scene reference (locks environment) |
| @Image 3 | Object / product | Product reference (locks object details) |
| @Video 1 | Motion / camera | Motion reference (locks action / camera) |
| @Audio 1 | Voice / rhythm | Voice reference (locks tone / timing) |
3. Shot Breakdown
Write your prompt in time order. Each shot should describe what the camera sees, who is speaking, and any sound.
0-3s: Wide shot, a man enters a parking garage. 3-6s: Medium shot, he opens the car door. 6-10s: Close-up, he starts the engine, camera slowly pushes in. 10-12s: He drives away, ambient city sound.
4. Text to Video
When no image is uploaded, describe the subject, scene, action and style in more detail.
A cat walking in a street, cinematic.
An orange cat walks slowly along a quiet city street, looking around curiously. The camera tracks backward in front of the cat, medium shot. Warm evening light, soft shadows, natural motion.
5. Image to Video
Don’t re-describe what’s already visible in the image. Focus on motion, camera behavior, and what happens next.
A woman in a red dress standing in a room.
The woman slowly turns her head toward the camera, her hair shifting in a soft breeze. The camera slowly pushes in, medium shot. Warm evening light.
6. Multi-Image & Audio
With multiple inputs, clearly assign a role to each file, then describe how they combine.
@Image 1: main character reference. @Image 2: product reference. @Audio 1: background music. A woman walks into a bright room, holding @Image 2's product. She smiles at the camera, warm natural lighting, medium shot. Camera slowly pushes in.
7. Camera & Motion
Describe camera movement and subject motion using clear, physical language.
Camera
slow dolly in dolly out tracking backward truck left + pan right handheld slow push-inSubject motion
walks slowly turns toward camera looks around raises one hand wind moves hair8. Editing & Refinement
To modify an existing generation, describe exactly what to keep and what to change.
Keep the character, action and camera movement unchanged. Replace the background with a modern office. Match the lighting direction to the original.
9. Common Mistakes
Break into Reference + Core Idea + Shot Description.
Always say what each file is for: character, scene, motion.
Write what the camera sees, not what it “feels like”.
Match dialogue length to shot duration.
10. Advanced Examples
Full prompts for more complex scenarios.
10s, 16:9 trap MV. Two performers in a dim underground space, hard cuts on the beat. Large block text appears with each bass hit. Gritty film grain, high contrast, no logos, no brand names.
15s, 16:9. A live-action train interior combined with hand-drawn glowing animation. Warm orange lines transform into shapes as the camera follows. No subtitles, no music.