Nano Banana Pro
Generate and edit images using Google's Gemini 3 Pro Image API, supporting text-to-image and image-to-image workflows at 1K, 2K, or 4K resolution.
by
Peter Steinberger
Nano Banana Pro Overview
Nano Banana Pro generates new images from text prompts or edits existing images with natural language instructions using the Gemini 3 Pro Image API.
It supports 1K, 2K, and 4K output sizes and a draft-to-final workflow so you can iterate fast at low resolution before generating a high-res final.
How It Works
Choose text-to-image or image-to-image, then pass the prompt and optional input image.
Invoke the Python script using `uv run` to call the Gemini 3 Pro Image API.
Generate 1K drafts to refine the prompt quickly.
When satisfied, output a 4K final for production-ready quality.
Key Features
Text-to-Image Generation
Create images from descriptive prompts with Gemini 3 Pro Image.
Image-to-Image Editing
Edit existing images with natural language instructions.
Multi-Resolution Output
Choose 1K (~1024px), 2K (~2048px), or 4K (~4096px) output sizes.
Draft-to-Final Workflow
Iterate quickly at low resolution, then generate a high-res final.
Timestamped Filenames
Outputs files with `yyyy-mm-dd-hh-mm-ss-name.png` naming.
Prompt Templates
Built-in templates for both generation and precise editing tasks.
Requirements
- No VM required.
- LLM API Key: set `GEMINI_API_KEY` for access to Gemini 3 Pro Image API.
- Python runtime: `uv` is required to run the script.
Use Cases
Concept Art & Mockups
Generate visual concepts for design projects in minutes.
Photo Editing
Remove objects, change backgrounds, or adjust styles with image-to-image.
Content Creation
Create illustrations and graphics for blogs, presentations, or social media.
Iterative Design
Refine prompts at 1K/2K before committing to a 4K final.
How to Install
npx clawhub@latest install nano-banana-proRun in your terminal, or click the Install button at the top of this page for one-click setup.
FAQ
What resolutions are supported?
The skill supports 1K, 2K, and 4K outputs.
Can I edit an existing image?
Yes. Provide an input image and editing instructions for image-to-image workflows.
What happens if I don't have a GEMINI_API_KEY set?
The API call will fail until you set `GEMINI_API_KEY`.
Where are images saved?
Outputs are saved to your current working directory with timestamped filenames.
Is there a recommended workflow?
Start with 1K drafts for speed, then produce 4K finals once the prompt is locked.