MiniMax H3 Max image-to-video animates an opening frame, with an optional closing frame, from a prompt describing the motion between them. Clips run 5 to 15 seconds at 480p or 768p, with ambience, foley and music generated alongside the picture.
Filtering differs by endpoint: Image to video — Softened, Video extension — Softened, Text to video — Filtered.
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.
- Output8s
grimoire-to-bare-shoulder A single unbroken tilt from an object up to a person, which makes the order of reveal the whole brief: what the camera finds first, and what it finds after. `image_url` fixes the opening composition so none of those words are spent describing a frame you already have. `duration_seconds` is `8` at the default `768p`, and `last_image_url` is there if you want the tilt to stop somewhere specific.
- 768p
- 8s
- Output5s
laundromat-spin Five seconds, the shortest this endpoint takes, from one frame and a plain movement note: the machine goes into its spin, she bounces with it, swings her legs and glances at the drum. Nothing about the room, the wardrobe or the flat fluorescent light is written down — the opening frame settled all of it.
- 768p
- 5s
- Output8s
frame-to-frame Both ends are pinned by stills: `image_url` sets her backlit and small at the far end of the colonnade, `last_image_url` sets the close stop-and-stare that ends the walk. The `prompt` only carries the eight seconds between them — the approach, the gown rippling with each step. Swap the closing frame for a different pose and the walk still lands clean at `duration_seconds` `8`.
- 768p
- 8s
This example starts at $0.05 per output second. The Playground updates the estimate when you change the settings.