Video Generation

Asynchronous AI video generation from text prompts or images is handled by creating a task and polling its status until a video URL is returned. This non-blocking approach allows applications to continue running while the video renders.

Image-to-video generation supports base64 strings or image URLs as input. You submit the image with your configuration parameters, start the task, and poll the status until the final video URL is retrieved.

AI video generation parameters include configurable settings for video quality, size, frames per second (fps), and duration. These options allow you to tailor the output to meet different production needs like marketing previews or tutorials.

Batch video generation is supported using the z-ai-web-dev-sdk. You can automate the creation of multiple videos at scale by submitting numerous text-to-video or image-to-video tasks and polling each one asynchronously.

The best way to monitor asynchronous video generation tasks is through task polling. After submitting your prompt or image, your application periodically checks the task status until the generation completes and provides the final video URL.

Asynchronous tasks prevent blocking the user interface during AI video generation. Because rendering takes time, polling a task ensures your application remains responsive while waiting for the final video URL to be available.