Image Generation: Create images with Gemini

Overview

Generate high-quality images using Google Gemini’s image generation models. Supports text-to-image generation and can optionally use existing images or videos as reference.

How to use

bash ${CLAUDE_PLUGIN_ROOT}/scripts/gemini.sh --model=gemini-3-pro-image-preview "YOUR IMAGE DESCRIPTION"

Arguments:

• --model – Required for image generation: Use gemini-3-pro-image-preview
• --file – Optional: Reference image or video for style/content guidance
  • When relevant, explicitly ask the model to include or preserve the original image/composition in the output.

Examples:

# Generate an image from a text description
npx -y superconductor-gemini-skills --model=gemini-3-pro-image-preview "A serene mountain landscape at sunset with golden light"

# Generate an image inspired by an existing image
npx -y superconductor-gemini-skills --model=gemini-3-pro-image-preview --file=reference.jpg "Transform this into a watercolor painting"

# Keep the original composition while applying a new style
npx -y superconductor-gemini-skills --model=gemini-3-pro-image-preview --file=reference.jpg "Keep the original image composition and main subjects, but restyle it as a polished dark-mode mobile app mockup"

# Generate an image from a video
npx -y superconductor-gemini-skills --model=gemini-3-pro-image-preview --file=video.mp4 "Generate an image that captures the essence of this video"

# Create specific styles
npx -y superconductor-gemini-skills --model=gemini-3-pro-image-preview "A cyberpunk cityscape in neon colors, digital art style"

Output

Generated images are saved to the current directory as gemini-image-{timestamp}.png.

API Key

The GEMINI_API_KEY environment variable must be set. Get your key at: https://ai.google.dev/gemini-api/docs/api-key

Models