Skip to main content

Alibaba's Wan 3.0 image-to-video starts from an opening frame and directs 2 to 30 seconds of motion with generated audio. An optional closing frame pins where the shot lands, at 480p, 720p or 1080p.

From $0.045/s10% below the publisher's list price

Workbench

These examples were written for this endpoint. Replace the prompt and sample media URLs with your own accessible inputs, then check the current quote and model availability before submitting.

  • Output5s
    yacht-fireworks-turn

    `image_url` and `last_image_url` are two stills of the same woman in the same silver sequin gown on the same yacht bow, shot from one camera position: in the first she stands in profile watching the harbor as a single rocket climbs; in the last she faces the lens, both hands on the rail, under a sky full of fireworks. With both ends pinned, the prompt only describes what happens in between: over `duration_seconds` `5` the fireworks build, she turns from the water, draws one fingertip down the neckline of her gown and settles against the rail, landing on the second still. `720p` with `generate_audio` on.

    • 720p
    • audio
    • 5s
  • Output5s
    rooftop-dip

    `image_url` and `last_image_url` are two stills of the same couple on the same rooftop from the same camera position: an open dance hold to start, the bottom of a dip to end. With both ends fixed the prompt only has to describe the move in between, and the clip opens on the first still and lands on the second with the dress, the shirt and the string lights unchanged. `duration_seconds` `5` at `720p` with `generate_audio` on.

    • 720p
    • audio
    • 5s
  • Output5s
    lingerie-window-turn

    The literal path for styled shots: the opening frame already carries the wardrobe, so the prompt only describes the turn.

    • 1080p
    • audio
    • 5s
  • Output5s
    bottle-hero-return

    `image_url` and `last_image_url` are the same file, so the rotation reverses and lands exactly where it started — a boomerang loop with no stitching in post. At `resolution` `1080p`, `duration_seconds` `5`, the highlight sweeps the glass once and returns as the bottle settles back; the label stays centred rather than drifting. `seed` `20260907` keeps it reproducible.

    • 20260907
    • 1080p
    • 5s

This example starts at $0.045 per output second. The Playground updates the estimate when you change the settings.