Video Generation

To generate AI videos from text prompts, use the z-ai-web-dev-sdk in your backend to create a task with a prompt, then poll asynchronously until the result returns a video URL. You can configure size, fps, duration, and optional audio.

Yes, you can create videos from base64-encoded images or image URLs. The image-to-video generation process accepts a single base64 string or a URL as input to guide the video creation workflow.

Asynchronous polling for video generation works by creating a task via the SDK and repeatedly checking its status until completion. The system returns a video URL in one of multiple potential fields when the task finishes processing.

Yes, you need the z-ai-web-dev-sdk installed in your backend code to run video generation. This dependency handles the task creation, asynchronous polling, and retrieval of the final video URL.

For marketing, education, and entertainment workflows, the video generation supports configurable output settings including video size, frame rate (fps), duration, and optional audio to meet specific content requirements.

Asynchronous video generation tasks return multiple potential URL fields to handle varying API responses. You should check these fields when polling completes to locate the generated video URL for your pipeline.