Image-to-Video Prompt Guide: Motion Without Losing the Subject

Image-to-video generation starts with a visual decision that has already been made. The subject, composition, colors, and much of the lighting exist in the source image. A useful prompt should therefore spend less time describing what is already visible and more time directing how the shot changes.

The five-part motion brief

Use this structure for most image-to-video prompts:

1. Preserve — identify what must remain stable.
2. Subject motion — describe one clear action.
3. Camera motion — choose one primary camera move.
4. Environment motion — add restrained secondary movement.
5. Ending — define how the shot settles or transitions.

For example:

Preserve the product shape, label, and studio lighting. The bottle rotates slowly by fifteen degrees while a narrow highlight travels across the glass. The camera makes a gentle push-in. Fine mist drifts in the background. End on a centered hero frame with the label fully readable.

Say what must not drift

If identity matters, name the stable elements before describing motion:

• facial identity and hairstyle
• product geometry and logo placement
• wardrobe and accessories
• background architecture
• text and packaging layout

Avoid asking for several large transformations at once. A dramatic subject action, fast camera orbit, changing environment, and wardrobe transition in the same short clip create competing constraints.

Choose one camera intention

Camera language should be specific but not overloaded:

• slow push-in
• gentle pull-back
• left-to-right tracking shot
• subtle handheld drift
• locked camera with subject motion only

If the source image already has a strong composition, a restrained camera move usually preserves it better than a full orbit.

Use environmental motion as support

Secondary movement makes a static image feel alive without changing the main subject. Useful examples include:

• fabric or hair moving in a light breeze
• steam, smoke, dust, rain, or mist
• reflections and moving highlights
• background pedestrians in soft focus
• leaves, water, curtains, or particles

Keep these elements subordinate to the main action.

Match the model to the shot

The available models expose different resolutions, durations, reference limits, and audio behavior. Choose the model from the composer first, then work within the controls shown for that model. A longer duration is not automatically better; short clips are often easier to keep coherent when the source image must stay recognizable.

Iterate with targeted changes

After the first take, avoid rewriting the entire prompt. Continue in the same chat and describe the smallest useful adjustment:

• “Keep the motion, but reduce the camera push-in by half.”
• “The product shape is correct. Slow the rotation and remove the background mist.”
• “Preserve the first three seconds, then hold the final frame longer.”

This makes it clear which parts of the previous result should survive the next generation.

Final checklist

Before generating, confirm that the prompt answers these questions:

• What must remain unchanged?
• What is the single main action?
• How does the camera move?
• Which secondary details move?
• What should the final frame look like?
• Is the requested motion realistic for the selected duration?

A good image-to-video prompt is a motion brief, not a second description of the still image.