nano-banana

AI-powered image, diagram, and video generation for Claude Code
Using Google Gemini API and Veo 3.1 with intelligent quality review

---

Features

• Style Presets - --style technical|visual-abstract|minimal via system_instruction — aesthetics separated from content
• Multi-Turn Chat - Iterative refinement retains context via client.chats.create()
• google-genai SDK - Single SDK for all Gemini and Veo models
• Video Generation - Veo 3.1 text-to-video, image-to-video, frame interpolation
• Smart Iteration - Only regenerates when quality is below threshold (saves API calls)
• Document-Type Aware - 13 quality thresholds (journal, architecture, presentation, etc.)
• AI Quality Review - Gemini 3.1 Pro reviews each diagram generation
• Five Skills - Technical diagrams, visual abstracts, general images, video generation, and text-based rendering
• Visual Abstracts - Nature-quality scientific figures with visual metaphors and isometric depth
• Image Editing - Modify existing images with natural language

Quick Start

1. Install the Plugin

git clone https://github.com/flight505/nano-banana.git

2. Install Dependencies

uv sync  # or: pip install google-genai

3. Configure API Key

Get a key at Google AI Studio:

export GEMINI_API_KEY='your-gemini-key-here'

Or use a .env file in your project:

echo "GEMINI_API_KEY=your-key-here" > .env

4. Generate!

# Technical diagram with quality review
python3 skills/diagram/scripts/generate_diagram.py \
    "Microservices architecture with API gateway" \
    -o architecture.png \
    --doc-type architecture

# Creative image
python3 skills/image/scripts/generate_image.py \
    "A cozy coffee shop on a rainy day" \
    -o coffee_shop.png

# Edit an existing image
python3 skills/image/scripts/generate_image.py \
    "Add rain to the window" \
    --input coffee_shop.png -o rainy_coffee_shop.png

# Generate a video
python3 skills/video/scripts/generate_video.py \
    "A time-lapse of clouds over a mountain" \
    -o clouds.mp4

# Image-to-video (animate a still image)
python3 skills/video/scripts/generate_video.py \
    "Slowly pan across the scene" \
    --input landscape.png -o animated.mp4

Skills

Diagram Skill

Generate publication-quality technical diagrams with AI quality review and style presets.

# Standard technical diagram
python3 skills/diagram/scripts/generate_diagram.py "User authentication flow" -o auth.png --doc-type architecture

# Visual abstract with dark background and glow
python3 skills/diagram/scripts/generate_diagram.py "API gateway as routing prism" -o visual.png --style visual-abstract --doc-type journal

# Wide diagram with aspect ratio
python3 skills/diagram/scripts/generate_diagram.py "System overview" -o overview.png --aspect-ratio 16:9

Document Types:

Type	Threshold	Use Case
specification	8.5/10	Technical specs, PRDs
architecture	8.0/10	System architecture
journal	8.5/10	Academic papers
presentation	6.5/10	Slides (faster)
readme	7.0/10	Documentation

Full Diagram Documentation

Image Skill

Generate and edit images using various AI models.

# Generate
python3 skills/image/scripts/generate_image.py "Abstract art in blue and gold" -o art.png

# Edit existing image
python3 skills/image/scripts/generate_image.py "Make the sky purple" --input photo.jpg -o edited.png

Available Models:

• gemini-3.1-flash-image-preview (default — Nano Banana 2, fastest)
• gemini-3-pro-image-preview (Nano Banana Pro, highest quality)

Aspect Ratio & Resolution:

# Generate with specific aspect ratio and resolution
python3 skills/image/scripts/generate_image.py \
    "A wide cinematic landscape" -o landscape.png \
    --aspect-ratio 16:9 --resolution 2K

Full Image Documentation

Kroki Skill

Render text-based diagrams (Mermaid, PlantUML, GraphViz, D2, and 23 more) to PNG/SVG.

# Render Mermaid to PNG
python3 skills/kroki/scripts/render_diagram.py -t mermaid -o flow.png \
    --source 'flowchart LR; A-->B-->C'