Generate Video

Generate videos using Google Veo models via @the-focus-ai/nano-banana.

Requirements

• npx must be installed.
• GEMINI_API_KEY environment variable must be set.

Usage

npx @the-focus-ai/nano-banana --video "prompt" --output "output_path" [options]

Prompting Best Practices (Google Veo)

Treat the prompt like a screenplay or shot list. Act as a Director of Photography.

Structure: Shot Composition/Camera Movement → Subject → Action → Setting → Aesthetics → Audio

Key Elements:

• Camera Movement: Use specific terms like dolly in, tracking shot, pan right, truck left, handheld, locked-off.
• Subject & Action: Describe what the subject is doing clearly.
• Aesthetics: "Cinematic", "35mm film", "grainy", "vibrant colors".
• Audio: Veo generates audio by default. You can prompt for specific sounds or dialogue (e.g., Sound of rain, Character says: "Hello world").

Workflow:

• Use Fast Mode (--video-fast) for testing movements and composition (cheaper).
• Use Standard Mode for final high-quality renders.
• Provide a Reference Image (--reference) ensuring character consistency.
• Provide an Input Image (--file) to animate a static image (Image-to-Video).

Parameters

• prompt (string, required): The description of the video.
• output_path (string, required): File path to save the generated video. Use --output "path/to/video.mp4".
• input_image (string, optional): Path to a static image to animate (Image-to-Video). Use --file "path/to/image.png".
• reference_image (string, optional): Path to a reference image for character consistency (max 3). Use --reference "path/to/ref.png".
• duration (integer, optional): Duration in seconds (4, 6, or 8). Default: 8. Use --duration 8.
• resolution (string, optional): "1080p" or "720p". Default: "1080p". Use --resolution 720p.
  • Note: 1080p requires 8s duration. 720p allows 4, 6, or 8s.
• aspect (string, optional): "16:9" (landscape) or "9:16" (portrait). Default: "16:9". Use --aspect 16:9.
• video-fast (boolean, optional): Use the faster, cheaper Veo model. Use --video-fast.
• no_audio (boolean, optional): Disable audio generation (saves cost). Use --no-audio.

Examples

Image-to-Video (Animate an image)

Use the --file flag to take a start image and animate it based on the prompt.

npx @the-focus-ai/nano-banana --video "The character turns and smiles" --file "images/character.jpg" --output "videos/smile.mp4"

Basic Video Generation

npx @the-focus-ai/nano-banana --video "Cinematic drone shot flying over a lush tropical island, golden hour lighting" --output "videos/island.mp4"

Fast Mode with Specific Settings

Use --video-fast for cheaper iterations.

npx @the-focus-ai/nano-banana --video "A robot dancing in a neon room" --video-fast --duration 4 --resolution 720p --output "videos/robot_dance.mp4"

Using a Reference Image

Use --reference to guide the style or character while generating a new scene.

npx @the-focus-ai/nano-banana --video "The character turns and smiles at the camera" --reference "images/character.jpg" --output "videos/smile.mp4"

Specific Camera Movement and Audio

npx @the-focus-ai/nano-banana --video "Close up, dolly in on a ticking antique clock. Sound of loud ticking." --output "videos/clock.mp4"