Alibaba Video API from prompts, frames and references
Explore Alibaba Video API guides for Wan and HappyHorse workflows. Choose text, image or reference inputs and follow the matching SeeAPI task flow.
Working with the Alibaba Video API
Choose how the clip begins
A written scene, a starting frame and a set of references represent different workflows. Select the Wan or HappyHorse entry matching the material your application already has.
Specify motion as well as appearance
Describe what changes through the shot, including subject movement and camera intent. Check the endpoint for duration, aspect ratio and any supported reference controls.
Retrieve the completed clip
Implement submission and the documented task-status flow together. Keep the result retrieval step in your integration so a task identifier is not mistaken for the finished video.
Alibaba Video API: practical questions
Should I use text-to-video or image-to-video?
Use text when the scene starts from a description; use an image workflow when an existing frame should guide the clip. Follow the selected endpoint’s requirements.
Are reference inputs the same across Wan versions?
Consult each version’s guide. Reference-to-video, first-frame and other input modes should not be treated as interchangeable request bodies.
How does HappyHorse fit this directory?
It has its own model and task entries. Compare those documents with the relevant Wan guide rather than assuming both families accept the same options.
What should an AI assistant receive for this workflow?
Provide one endpoint’s Markdown and your source assets, target clip requirements and task-handling needs. That keeps the generated code tied to a specific contract.