← All guides

The best image to video AI for your photo

A printed photograph on a table whose subject seems to step out of the frame in motion blur.

Turning a photo into a moving shot: which AI model holds your image best, steers the ending, or tests ideas cheaply.

1 min read · last checked

The frame follows your image

Image to video starts from your picture, and with most models the shape of the clip comes from that picture rather than from a setting. Crop the still to the shape you want before you generate, because that decision is made before the model runs.

The best result from one photo: Kling 3.0

Kling 3.0 Pro is the top tier of the Kling family for image to video: cinematic visuals, fluid motion and sound in the same pass, up to fifteen seconds.

Deciding where the shot ends: Wan 3.0 or MiniMax H3

Give the model both ends of the movement and it fills in the middle. Wan 3.0 takes a start and an end frame and makes its own audio. MiniMax H3 takes a first and last frame too, and drafts cheaply at 480p before a final at up to 4K.

Testing ideas cheaply

When you are still finding the shot, a cheap model is the right tool.

  • Vidu, with cheap 360p drafts and 4K with sound at the top.
  • Pika 2.2, which turns a still into a shot with camera and subject moving together.
  • PixVerse V6, with four resolutions and sound when you want it.

Read next

Kling 3.0The strongest image to video here.Wan 3.0Start and end frames, with its own audio.Start and end framesWhen two frames beat a prompt.ViduCheap drafts, 4K with sound at the top.Kling image to videoHow to prompt it from a photo.

Put this guide to work

The studio opens in stages. We open access in the order people signed up, and you start with 1,000 free credits.

Get early access