fal-veo3-image-to-video

Required inputs

• A text prompt describing the desired motion/scene.
• One reference image (--image) as a local file or URL.
• A fal.ai API key in workers.jsonc (fal.key).

Optional inputs

• Additional model parameters via --param k=v (repeatable).
  • Value coercions:
    • true / false
    • numbers (e.g. 24, 0.7)
    • json:<JSON> (e.g. json:{"a":1})
    • @path/to/file.json (read JSON from file)
  • Nested keys are supported via dot notation (e.g. --param camera.motion=fast).
• --sync-mode to call fal.run directly instead of queue polling.
• --image-field to force the request field name for images (defaults to image_url for 1 image and image_urls for >1).