Optional modelModel identifier to use for generation.
Source image (the first frame): raw bytes, an https:// URL or a data:
URL. See VideoImageInput for how each provider sends it. A custom
provider that sends a URL can pass it through videoImageToUrl() (from
the video index), which leaves a string as it is and turns a Buffer into
a data: URL. A string reaches a provider only when the caller passed one.
Text prompt describing the desired motion / narrative.
Optional negativeNegative prompt describing content to avoid.
Optional durationDesired output duration in seconds.
Optional aspectDesired aspect ratio.
Optional resolutionDesired output resolution, in the model's own vocabulary (for example
'720p', '1080p', or fal's '768P' for MiniMax). Sent as given.
Optional referenceImages of subjects or assets the video must keep consistent, besides the
source image. Sent as fal's reference_image_urls, Replicate's
reference_images, or Gemini image parts after the source image.
Optional outputOutput container format. Defaults to 'mp4'.
Optional fpsFrames per second for the output video.
Optional seedSeed for reproducible output.
Optional userIdentifier of the requesting user.
Optional providerProvider-specific model inputs: a VideoProviderOptionBag keyed by provider id, or a flat object that fal, Replicate and Gemini receive whole.
Request payload for image-to-video generation.
Passed to IVideoGenerator.imageToVideo by the high-level orchestration layer. Requires a source image that serves as the first frame (or style reference) for the generated video.