genimage

Installation
SKILL.md

Nano Banana — Image Generation

Generate and edit images via the Gemini API. If the API key is missing, run /nano-banana:setup.

Workflow

  1. Determine the mode from the user's request:

    • Creating a new image from text? → Text-to-image (no --images)
    • Editing/retouching an existing image? → Image editing (--images source.png)
    • Applying one image's style to another? → Style transfer (--images style.png target.png)
    • Combining or referencing multiple images? → Multi-image (--images a.png b.png ..., up to 14)
    • Need 2K or 4K output? → Add --resolution 2K|4K to any mode above
  2. Craft the prompt — describe the desired result narratively. See references/prompting-guide.md for domain-specific patterns (website visuals, placeholders, product shots, etc.).

  3. Run the script:

python "$CLAUDE_PLUGIN_ROOT/scripts/genimage.py" --prompt "..." [options]
  1. Show the result — read the output image and present it to the user.

Flags

Flag Required Purpose
--prompt "text" Yes Describe the desired image
--output file.png No Output path (default: generated_image.png)
--images path [...] No Input image(s) — omit for text-to-image
--aspect-ratio RATIO No 1:1 2:3 3:2 3:4 4:3 4:5 5:4 9:16 16:9 21:9
--resolution RES No 1K 2K 4K (uppercase K required, triggers Pro model)

Models

  • Nano Banana 2 (gemini-3.1-flash-image-preview): Default. Fast, high-quality.
  • Nano Banana Pro (gemini-3-pro-image-preview): Auto-selected when --resolution is set. 4K output, advanced reasoning, high-fidelity text rendering.

Common Use Cases → Flags

Use case Example command
Website hero image --prompt "..." --aspect-ratio 16:9
Social media post --prompt "..." --aspect-ratio 1:1
Mobile banner --prompt "..." --aspect-ratio 9:16
Placeholder / mockup --prompt "..." --output placeholder.png
Edit existing photo --prompt "remove the background" --images photo.png
Style transfer --prompt "Apply watercolor style" --images style.png photo.png
High-res asset --prompt "..." --resolution 4K

Editing

All editing is text-guided — describe what to change ("replace the sky", "remove the person", "add a sunset") and Gemini identifies and modifies regions automatically. No masks needed.

Installs
1
GitHub Stars
3
First Seen
Apr 15, 2026