How to Write an AI Video Motion Prompt
Updated: Nov 5
A useful video prompt describes what happens over time. Start with one visible action, then state how the camera observes it. Keep the production goal, exact titles and payment settings separate from the scene description.
Build the sentence in layers
Write subject and action first: a folded linen napkin gently lifts at one corner. Add the camera: a slow push-in from a tabletop view. Finish with the environment: soft window light and a quiet kitchen. This gives each phrase a job rather than stacking unrelated style words.
Example prompt
“A folded linen napkin rests beside a ceramic plate. One corner moves lightly in the breeze. The camera slowly pushes closer at table height. Warm daylight falls from the left. The scene has a calm, realistic appearance.” This is an illustrative prompt, not a tested output or a promise of exact motion.
Improve one variable at a time
If the camera moves too much, reduce the requested move before rewriting the entire scene. If the subject changes, simplify its action or improve the reference. Save the first prompt and the change you made so you can tell which instruction affected the result.
Write an image-to-video prompt for a slow camera push-in
Specify the subject and one controlled move, such as a slow push toward the product label while the product remains still. Check the middle of the shot for warped packaging before adding extra motion.
Change one variable between test prompts
Keep the subject and starting frame fixed, then change either the camera move or the subject action. Compare the results to see which instruction caused the change before combining movements.
Common confusion
“Cinematic, stunning, epic” does not specify an action. A detailed story can also be too much for a single shot. Break separate events into a plan and ask for a focused shot prompt for each. The chosen generator may interpret the same phrase differently.
Use the planner
In Create Video, describe the audience, source materials and desired result. Review the resulting plan and instructions before moving to generation. Keep any exact on-screen wording available for the finishing stage.
Fix a prompt that gives the camera two jobs
If your reference is a close photo of a perfume bottle, try: “The bottle stays on the table. The camera makes a slow forward move. The label stays facing the camera.” Review the framing and label before asking for a turn or a reveal. If the move starts too wide, keep the subject instruction and shorten the requested move instead of adding a second camera direction.
Put your plan into action
Bring your idea and reference photos to the video planner. Review the plan and available generation settings, then check the price before checkout. Generate and download the result on AI LEMNISCATA.
Explore this next
See how these motion instructions work in our step-by-step AI dog video example. The copyable idea describes the dog standing, sitting when the vendor turns, and returning to the counter within one continuous shot.