| name | nano-banana-pro |
| description | AI image generation using Google Gemini's gemini-3-pro-image-preview model. Simple script to generate images from text prompts. Use when user asks to generate, create, or draw images with Gemini. |
Image Generation with Gemini
Generate images using Google's Gemini model (gemini-3-pro-image-preview).
Script Directory
Agent Execution:
SKILL_DIR = this SKILL.md file's directory- Script path =
${SKILL_DIR}/scripts/generate_image.py
Prerequisites
- Python 3.8+
- Google API key with Gemini access
google-genai package installed
Installation
pip install google-genai
Environment Variables
| Variable | Description |
|---|
GOOGLE_API_KEY | Google API key (required) |
Usage
python ${SKILL_DIR}/scripts/generate_image.py --prompt "A serene mountain landscape" --output "output.png"
python ${SKILL_DIR}/scripts/generate_image.py --prompt "A cute cat" --output "cat.png" --model "gemini-3-pro-image-preview"
python ${SKILL_DIR}/scripts/generate_image.py --prompt "A futuristic city" --output "city.png" --temperature 0.8
Options
| Option | Description |
|---|
--prompt <text> | Prompt text for image generation (required) |
--output <path> | Output image filename (default: output.png) |
--model <id> | Gemini image model (default: gemini-3-pro-image-preview) |
--temperature <float> | Sampling temperature (default: 1.0) |
How It Works
- Takes a text prompt as input
- Sends the prompt to Gemini's image generation API
- Streams the response and extracts inline image data
- Saves the generated image to the specified output file
Example Prompts
- "A serene mountain landscape at sunset"
- "A cute cartoon cat playing with yarn"
- "A futuristic cityscape with flying cars"
- "A watercolor painting of a flower garden"
- "A minimalist logo design for a coffee shop"
Error Handling
- If no API key is set, the script will fail with an authentication error
- If image generation fails, the script outputs "็ๆๅพ็ๅคฑ่ดฅ"
- On success, outputs "ๅพ็ๅทฒไฟๅญๅฐ {output_path}"
Notes
- Uses streaming API for efficient response handling
- Supports PNG output format
- Model must support IMAGE response modality
