Skip to main content

Overview

Decart video models support standard AI SDK video generation settings with some limitations. This page describes the supported and unsupported settings for all video models.

Supported Settings

prompt

string
Text description to guide video generation.
  • Required for lucy-pro-t2v (text-to-video)
  • Optional for image-to-video models (provides additional guidance)

seed

number
Random seed for reproducible video generation.Using the same seed with the same inputs will produce the same video.

resolution

'1280x720' | '640x480'
Video resolution.Only two resolutions are supported:
  • "1280x720" - 720p (default)
  • "640x480" - 480p
Other resolutions will produce a warning and be ignored.

aspectRatio

'16:9' | '9:16'
Aspect ratio of the generated video.Only two aspect ratios are supported:
  • "16:9" - Landscape orientation
  • "9:16" - Portrait orientation
Other aspect ratios will produce a warning and be ignored.

image

ImageInput
Source image for image-to-video generation.
  • Required for lucy-pro-i2v, lucy-dev-i2v, and lucy-motion
  • Not used for lucy-pro-t2v
Supports two input types:
  • URL reference: { type: 'url', url: string }
  • Binary data: { type: 'binary', data: Uint8Array | string, mediaType: string }

Unsupported Settings

The following settings are not supported and will produce warnings:

fps

number
Not supported. Frame rate cannot be configured.Specifying this parameter will produce a warning of type unsupported.

duration

number
Not supported. Video duration cannot be configured.Specifying this parameter will produce a warning of type unsupported.

Provider-Specific Options

Decart video models support additional options via the providerOptions.decart parameter. See Video Provider Options for details.

Standard Request Options

headers

Record<string, string>
Custom HTTP headers to include with the request.

abortSignal

AbortSignal
AbortSignal to cancel the request.Cancels both the initial submission and polling process.

Handling Warnings

When using unsupported settings, the model returns warnings in the response: