Turning an image into a video: step by step

2026-07-13

Bringing an old family photo to life, turning a product shot into an ad video, or making a portrait speak — all of it takes about a minute with image-to-video technology.

What is image-to-video?

You give the model one (sometimes several) images and describe the motion in text. The model keeps the composition, colors and subject of your picture and sets it in motion. The result: a 5-15 second video that starts from your image.

The best image-to-video models

• Hailuo 2.3 — #1 for human portraits: facial expressions and body motion look remarkably natural.
• Kling — the all-rounder: stable results for objects, animals and landscapes.
• Seedance 2.0 — accepts up to 9 images; ideal for character consistency.
• Veo 3.1 — when you need audio and top quality.

Choosing the right image

• High quality — a blurry image produces a blurry video.
• Clear subject — the model needs to know what to animate.
• Simple background — busy backgrounds produce more artifacts.
• Matching aspect ratio — vertical image → 9:16 video.

How to write the motion prompt

In image-to-video the prompt should describe only the motion — not the visuals (they're already in the picture): "The girl smiles and looks at the camera, her hair sways in the wind" or "The camera slowly pushes forward as waves crash on the shore". Describe what the subject does, how the camera moves, and how the environment changes.

Upload an image and see for yourself — with the Remix feature you can also start from prompts of videos in the community gallery.