Image to Video Prompts: Describe the Motion, Not the Picture

The biggest difference between text-to-video and image-to-video is what your prompt has to do. With a photo, the model already knows the subject, colors and composition. Good image to video prompts therefore say almost nothing about how things look and almost everything about what moves: the subject, the environment and the camera. Redescribing the picture can even push the model to change it. Below you'll find a simple structure, examples by photo type and fixes for common problems. Upload an image at generatorvideo.ai, add a motion prompt, choose Turbo or Pro, and get a 3-30 second clip (free videos: 3, 5 or 8 s). The first 3 are free.

A three-layer structure for motion prompts

• Subject motion: what the main subject does. "She turns her head toward the camera and smiles."
• Environment motion: what moves around it. "Hair and scarf drift in a light breeze, leaves fall behind her."
• Camera motion: how the view changes. "Slow push-in."

Add a speed word to at least one layer: slowly, gently, suddenly. Example for a portrait: "She slowly turns toward the camera and smiles, strands of hair move in a soft breeze, slow push-in." That's it - no need to mention her red coat; it's in the image.

Keep the total amount of motion proportional to the clip length. A 3-second clip needs one small action; an 8-second clip can hold a slow camera move plus a subject gesture. Asking for a lot of movement in a short clip is the most common reason an animated photo looks rubbery or rushed.

Example prompts by photo type

Portraits and people

• "Blinks, then looks up and laughs softly, subtle head movement, static camera."
• "Steps forward toward the camera, coat moving with each step, camera slowly pulls back."

Products

• "Slow 360-degree orbit around the product, light glides across the surface."
• "Condensation drops run down the bottle, gentle push-in, background stays still."

Landscapes and travel

• "Clouds drift across the sky, water ripples, slow pan right."
• "Fog rolls through the valley, camera rises slowly like a drone."

Illustrations and art

• "Stars twinkle, the painted river flows, brushstroke texture preserved, very subtle motion."
• "The character's cape flutters, lanterns sway, slow zoom in."

Old family photos

• "Gentle breathing movement, slight smile, very slow push-in." Keep motion minimal for faces.

Choosing the right image

• Match the aspect preset to the photo. A tall portrait fits 9:16 or 4:5; a landscape fits 16:9. Mismatched shapes force cropping or odd extension.
• Sharp, well-lit sources animate better than blurry or tiny images.
• One clear subject is easier to animate than a crowded group photo.
• Leave room for motion. If the camera will push in or the subject will move, a bit of space around them helps.

More on the image workflow itself is on our image to video AI and photo to video AI pages.

Fixing common image-to-video problems

• Face changes or warps: reduce motion. Ask for small movements and a static or slow camera. Try Pro mode, which usually keeps faces steadier.
• Nothing moves: name specific moving elements ("steam rises", "leaves fall") rather than "make it alive".
• Product label distorts: avoid fast rotation; use a slow push-in or light moving across instead.
• Style drifts from the original: remove style words from the prompt; let the image define the look.
• Motion is too fast: add "slowly" and pick a longer clip length.

Turbo or Pro, and cost

Turbo is our fast option for testing how a photo reacts to motion. Pro is the quality option when faces, hands or fine texture matter. Both cost $1 per clip after your 3 free videos. Example: animating 10 product photos with one retry each is 20 clips, one $20 top-up. Clips appear in our public feed by default - Unpublish anything private, especially personal photos. Upload an image to start.