How to Write Image to Video Prompts

Treat the image as frame one

Do not spend most of the prompt redescribing information that is already visible. Use the prompt to define motion, timing, camera behavior, environmental response, and stability constraints.

Before generating, inspect the source frame. Motion will expose weak edges, unreadable details, awkward poses, and insufficient space around the subject. Start from a clean image with room for the intended move.

Write motion in layers

Separate motion into primary, secondary, and environmental layers. Primary motion is the main subject action. Secondary motion includes hair, clothing, reflections, or product mechanisms. Environmental motion includes rain, dust, foliage, traffic, or light.

Example: “The subject slowly turns toward camera; loose fabric trails with natural inertia; soft wind moves the grass in the background. Camera makes a restrained five-percent push-in. Maintain facial identity and wardrobe details.”

• Primary: the action viewers should notice first
• Secondary: small physical responses that sell realism
• Environment: motion that gives the shot atmosphere and depth
• Stability: details that should not transform

Choose one camera move

Short clips usually benefit from one readable camera action. Combine a push-in with a pan, orbit, crane, and zoom only when the shot genuinely needs all of them. Too many directions can produce unstable geometry.

Use practical language: locked camera, slow push-in, lateral track, gentle orbit, handheld drift, tilt up, or pull back. Add speed and intention rather than only naming the move.

Describe pacing and physical behavior

Words such as slowly, restrained, sudden, accelerating, or settling help define timing. For materials, describe physical behavior: silk floats and settles, rigid packaging stays fixed, liquid sloshes, and metal reflections move with the camera.

Avoid asking every object to move. A stable anchor gives the clip a clear visual hierarchy and makes the intended motion easier to read.

Protect identity and product details

State what must remain consistent: facial identity, garment construction, product label, silhouette, material, or background architecture. Use modest motion when exact visual accuracy matters.

Review frame by frame before commercial use. Motion can introduce errors that are not visible in the starting image, especially around text, hands, logos, and reflective surfaces.

What should I include in an image-to-video prompt?

Describe subject motion, camera movement, pacing, environmental motion, and the details that must remain stable.

Why does my image-to-video result change the subject?

Large motion, complex camera moves, or ambiguous source details can cause drift. Use a clearer image, reduce motion, and explicitly preserve important features.

Should I describe the source image again?

Only repeat details that must remain stable. Spend most of the prompt on motion and timing because the source image already supplies the visual composition.