Writing a Clear Video Prompt

A useful prompt is a short shot brief. In EveryGen AI, start with the scene you want to see and choose settings separately in the studio.

Subject and action

Name one main subject and a concrete action: a perfume bottle rotates, a runner turns a corner, or a paper lantern lifts into the air. Avoid combining several competing actions in the same short clip.

Scene and light

Describe the setting and light in a few specific words: a dark studio with soft side light, a rainy street at blue hour, or a warm window-lit room. A consistent environment gives the shot direction without an excessive adjective list.

Camera movement

Choose one camera instruction, such as a slow push-in, gentle orbit, locked-off shot, or low tracking view. Multiple conflicting movements can make the result harder to control.

Example: “A glass perfume bottle rotates slowly on a charcoal pedestal. Soft amber side light catches the glass. The camera makes a gentle close-up push-in; the background remains dark and simple.”

Working from an image

For a supported image-to-video model, describe the movement you want instead of rewriting everything visible in the reference. Use an authorised, publicly accessible reference image. Do not include confidential URLs or another person's sensitive information.

Iterate and inspect

Change one element per attempt so you can compare results. Keep model, duration, and ratio stable when comparing prompts, and check the credit cost before every submission. These techniques give a clearer brief; they do not guarantee a particular output.

Try it in the video studio. The cover is an original concept image, not proof of a model's video output. For commercial use, review permissions and our Terms of Service.