Video generation API for AI assistants

The video generation API takes a written description of a scene, and optionally a first frame, and returns a link to a short video clip, usually five or ten seconds long. Your assistant can make a draft clip for a post or a slide, and you decide whether to use it.

What it does

You pass a prompt, and you can set the length and the shape of the clip, e.g., wide for a slide or tall for a phone. Some models also take an image and animate it. The model renders the clip, and the reply has a link to the video file.

Rendering takes longer than a data call, from several seconds to a few minutes, and the price depends on the model and the length of the clip.

How a Joshu assistant uses it

When you write a launch post, Joshu can make two or three short clips from your description and put them in a folder on its desktop for you to pick from.

When you have a product photo, Joshu can turn it into a slow pan or a short loop for a social post, and keep the original next to it.

Choosing a provider

Each provider here is a different video model, and they differ more in the look of the result than in the facts. Some are better at people and motion, and others at text, products, or a steady camera.

Start with the cheaper model for drafts, and move to a more expensive one for the clip you will publish. Every call is charged whether you keep the clip or not, so ask Joshu for one draft at a time when the budget is tight.

curl https://hello.joshu.me/api/router/call \
  -H "Authorization: Bearer $JOSHU_ROUTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "endpoint": "auto",
    "capability": "video.generate",
    "params": {
      "prompt": "<prompt>"
    }
  }'