How to Prompt for AI Video Generation

A video prompt describes a shot that moves through time. Beyond how the scene looks, you describe what happens, how the camera moves, and what the viewer hears. Models such as Veo 3.1, Kling, Seedance, and Wan 3.0 respond well to director-style language.

Prompt a single shot, not a whole movie.

Shot type

Close-up, medium, wide, or aerial

Subject

Who or what, with the key details

Action

What happens, step by step

Setting

The place and the time of day

Camera

Movement and lens choice

Lighting and mood

The look and the feeling

Audio

Dialogue, sound effects, and ambience

Format

Duration and aspect ratio

• Give each shot one main action
• Lay out the sequence: first, then, finally
• Keep every movement physically possible
• Avoid crowds of people doing different things

A runner ties her shoe on a foggy bridge, stands up, watches the sunrise, then jogs away from the camera.

Static

Calm and observing

Slow push-in

Builds emotion or suspense

Tracking shot

Follows the subject closely

Pan or tilt

Reveals the surrounding space

Drone or crane

Conveys scale

Handheld

Adds realism and energy

• Put spoken lines in quotes
• State who speaks and how
• Add sound effects and background ambience
• Keep each shot’s dialogue short

Many current models, Veo 3.1 and Kling among them, generate sound along with the video.

Image-to-video gives you more control. Create the ideal first frame as an image, then animate it with a prompt about motion and camera, not about what the image already shows.

Steam rises slowly from the cup, a soft breeze stirs the curtain, slow push-in, natural and subtle movement.

• Set first and last frames to control where a shot begins and ends
• Use up to 10 keyframes along the timeline in Kling 4.0
• Add reference images for characters, products, and places
• Name each reference’s role in the prompt

• Reuse the same reference images in every shot
• Describe the character identically each time
• Hold wardrobe, lighting, and style steady
• Generate shots in story order and review each one

• Short shots are easier to control. Clips of 5 to 10 seconds work for most scenes
• Some models can now generate up to 30 seconds in a single pass
• Pick 9:16 for social, 16:9 for YouTube and film
• Cut many short shots together into a longer video

Medium shot of an elderly fisherman mending a net on a wooden pier at golden hour. He pauses, gazes out to sea, and smiles. Slow dolly-in, 50mm lens, shallow depth of field, warm amber light. Soft waves and distant gulls. He says quietly, “Good catch today.” Eight seconds, 16:9.

Warped faces or hands

Shorten the clip, cut the motion, or start from a clean image

Ignored camera moves

Place the camera instruction early and keep it simple

Too much going on

Split it into two separate shots

Inconsistent characters

Add reference images and reuse the same description

• Never make misleading videos of real people or events
• Use only faces you have permission to use
• Respect copyrighted characters, music, and brands
• Label realistic AI video wherever platforms require it