Vidu: video generator — what it does, formats and rights to the result

Vidu is a cloud-based video generator from Shengshu that works on the text-to-video and image-to-video models. The service lets you create short animated clips from a text description, a static image, or multiple angles of one object, as well as improve the quality of existing video.

What it does

The developer states four main generation modes:

• Text2Video — creating video from a text description using keyframes to maintain a consistent theme.
• Image2Video — turning a static image into dynamic video with animation and narrative development.
• Reference2Video — generating video based on a reference image and text description. Supports working with characters and objects; you can upload multiple angles of one item to preserve its visual consistency in the frame.
• Start-End2Video — converting static images into dynamic clips; the authors describe this mode in the same terms as Image2Video.

There is also a separate Upscale feature — increasing video resolution from 360p to 1080p.

Input is text, an image, or multiple images of an object. Export formats, as described by the author: png, jpeg, jpg, webp. Available video resolutions: 540p, 720p, 1080p. The maximum clip duration documented for the Vidu Q3-pro model is 16 seconds; other model versions may have different limits — from 5 to 16 seconds.

The API lists the following model versions: viduq3-pro-fast, viduq3-turbo, viduq3-pro, viduq2-pro-fast, viduq2-pro, viduq2-turbo, viduq1, viduq1-classic, vidu2.0.

How to use it

Vidu is a cloud service available via API. A web interface or standalone app is not mentioned in the fact sheet. The service does not release its own model or weights, and running on your own hardware is not possible. API documentation is collected at: https://platform.vidu.com/docs/llms.txt.

Rights to the output

The provided fact sheet does not describe the licensing terms or ownership of rights to generated videos. This should be clarified in the service's user agreement.

Who it suits

The service can be useful for those who need quick animation of static images or sketches, as well as creating short videos from text descriptions without deploying local infrastructure. The Reference2Video mode with multiple object angles is aimed at tasks where visual consistency of a character or item from frame to frame matters.

For those looking for an open source generator with the ability to run locally or fine-tune the model, Vidu is not suitable — it is a closed cloud API. Also keep in mind that the maximum clip duration is limited to 16 seconds, and work is done through a credit system, the details of which the authors disclose in the documentation.

Vidu covers basic scenarios of turning text and images into short video, but the details of how each mode works and the angle limitations for Reference2Video should be checked in the API documentation — they are not covered in the fact sheet.