Prompting for video is fundamentally different from prompting for images. You need to describe not just what things look like, but how they move, how the camera behaves, and how the scene evolves over time.

The Video Prompt Structure

[Scene description] + [Motion/Action] + [Camera movement] + [Mood/Lighting] + [Technical specs]

Example: "A woman in a red coat walks through a rainy Tokyo street at night, neon signs reflecting on wet pavement, slow tracking shot following from behind, cinematic, anamorphic lens flare, 24fps film grain"

Describing Motion

Be specific about how things move:

• "A bird takes off from a branch, wings spreading wide, slow motion"
• "Waves crash against rocks, water spraying upward, real-time speed"
• "A dancer spins gracefully, dress flowing outward, smooth continuous motion"

Camera Language

Use cinematography terms that AI models understand:

• Static shots — "locked-off camera", "tripod shot", "fixed frame"
• Tracking — "dolly forward", "tracking shot left to right", "steadicam follow"
• Aerial — "drone shot rising upward", "bird's eye view slowly rotating"
• Close-ups — "extreme close-up on eyes", "macro shot of dewdrops"
• Transitions — "slow zoom in", "pull focus from foreground to background"

Temporal Descriptions

Describe how the scene changes over time:

• "Starting at dawn, transitioning to golden hour"
• "Beginning with an empty room, then filling with people"
• "The flower blooms in time-lapse over the course of the clip"

Style and Quality Keywords

• "cinematic, 4K, shallow depth of field, film grain"
• "documentary style, handheld camera, natural lighting"
• "music video aesthetic, high contrast, vivid colors, fast cuts"
• "commercial quality, clean, bright, professional"

Common Mistakes

• Too many actions — Models handle one or two clear actions better than a complex sequence
• Ignoring physics — Describe physically plausible scenarios for better results
• No camera direction — Without camera guidance, results may be static or erratic
• Overloading the prompt — Keep it focused; describe the most important 5-10 seconds