Skip to content
Sup

sup.ai.video

The sup.ai.video package provides AI-powered video generation capabilities using Google’s Veo models. It is accessed through sup.ai.video.

Example sup.ai.video Usage
// Generate a video from text prompt
const video = sup.ai.video.create("a cat playing with a ball of yarn");


// Generate with custom options
const hdVideo = sup.ai.video.create(
  "sunset over ocean with waves",
  {
    duration: 8,
    aspectRatio: "16:9",
    resolution: "1080p",
    model: "best"
  }
);


// Generate from an image
const imageToVideo = sup.ai.video.create(
  sup.input.image,
  "animate this image with gentle movement"
);

Methods

sup.ai.video.create()

(prompt: string | SupImage, options?: VideoCreateOptions) → SupVideo

// Simple text-to-video
const video = sup.ai.video.create("a bird flying through clouds");


// With custom duration and aspect ratio
const video = sup.ai.video.create("dancing robot", {
  duration: 6,
  aspectRatio: "9:16",
  resolution: "720p"
});


// Image-to-video
const video = sup.ai.video.create(
  sup.input.image,
  "make this image come alive with subtle animation"
);

Creates a video using AI based on a text prompt, image, or existing video.

Parameters:

  • prompt (string | SupImage | SupVideo):
    • Text prompt: Description of the video to generate
    • SupImage: Starting image to animate
  • description (optional string): When using image or video input, provide a text description
  • options (optional object):
    • duration (number): Length of video in seconds (default varies by model)
    • aspectRatio (string): Video dimensions:
      • "16:9": Landscape (default)
      • "9:16": Portrait
    • resolution (string): Output quality:
      • "720p": HD quality (default)
      • "1080p": Full HD quality
    • model (string): Which Veo model to use:
      • "fast": Fast generation with Veo 3.1
      • "best": Highest quality with Veo 3.1 (default)
      • "veo-3.1-fast-generate-001": Explicit fast model
      • "veo-3.1-generate-001": Explicit best model
      • "veo-3.1-fast-generate-preview" / "veo-3.0-fast-generate-preview": Legacy aliases mapped to fast GA endpoint
      • "veo-3.1-generate-preview" / "veo-3.0-generate-preview" / "veo-2.0-generate-preview" / "veo-2.0-generate-exp": Legacy aliases mapped to best GA endpoint

Returns: A SupVideo object containing the generated video

Examples:

// Portrait video for social media
const socialVideo = sup.ai.video.create(
  "person dancing to upbeat music",
  {
    aspectRatio: "9:16",
    duration: 8,
    resolution: "1080p"
  }
);


// Quick preview with fast model
const quickPreview = sup.ai.video.create(
  "car driving down highway",
  {
    model: "fast",
    duration: 4
  }
);


// Animate a still image
const animatedPainting = sup.ai.video.create(
  artworkImage,
  "bring this painting to life with gentle movements and lighting changes"
);

sup.ai.video.interpret()

(...args: (string | SupVideo)[]) → string

const video = sup.input.video;
const description = sup.ai.video.interpret(video);


// With a custom prompt
const analysis = sup.ai.video.interpret(
  video,
  "What actions are happening in this video?"
);

Analyzes a video using AI and returns a text description. Uses Gemini 3 Flash for multimodal video understanding.

Parameters:

  • video (SupVideo): The video to analyze
  • prompt (optional string): Custom instructions for the AI analysis. If not provided, uses a default prompt.

Returns: A string containing the AI’s interpretation of the video

Examples:

// Basic video description
const video = sup.input.video;
const description = sup.ai.video.interpret(video);


// Analyze specific aspects
const actions = sup.ai.video.interpret(
  video,
  "Describe the main actions and movements in this video."
);


// Transcribe speech in video
const transcript = sup.ai.video.interpret(
  video,
  "Provide a transcript of all spoken words in this video."
);

Notes

  • Generated videos are returned as SupVideo objects which can be displayed by returning them from main().
  • The Veo models excel at understanding natural language descriptions and generating realistic motion.
  • Video generation can take longer than image generation depending on duration and quality settings.
  • For best results, provide detailed descriptions including camera movement, lighting, and action.
  • The "best" model produces higher quality but takes longer to generate than "fast".
  • Image-to-video can be used to animate still images with motion.