nano-banana-edit

Edit images via RunComfy CLI against the google/nano-banana-2/edit endpoint.

31|9|Updated Apr 30, 2026
One-click install
npx skills add https://github.com/agentspace-so/runcomfy-agent-skills --skill nano-banana-edit-agentspace-so
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: nano-banana-edit
Source: https://github.com/agentspace-so/runcomfy-agent-skills/tree/main/nano-banana-edit
Command: npx skills add https://github.com/agentspace-so/runcomfy-agent-skills --skill nano-banana-edit-agentspace-so

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you edit existing images (image-to-image) while preserving the subject’s identity, so you can reliably swap backgrounds or make localized changes without re-creating the whole image.

Core Features & Use Cases

  • Identity-preserving background and clothing edits: Maintain face/pose/clothing integrity while changing the scene.
  • Localized, spatial edits: Target specific regions like “left object only” or “upper-right corner.”
  • Up to 20-image batch workflows: Apply consistent edits across multiple input images, with locked framing via aspect ratio and resolution.
  • Route to the right edit sibling when needed: Use GPT Image 2 edit for in-image text, Flux Kontext for precise single-reference local edits, and Nano Banana 2 t2i for generation from scratch.
  • RunComfy CLI execution: Runs google/nano-banana-2/edit through the local RunComfy CLI and downloads results to an output directory.

Quick Start

Use the nano-banana-edit skill to replace the background of a portrait image while keeping the subject’s identity unchanged.

Frequently Asked Questions about nano-banana-edit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I edit an image and swap backgrounds while keeping the subject identity unchanged?

To edit an image and preserve subject identity during a background swap, use the nano-banana-edit Skill. It runs the google/nano-banana-2/edit endpoint to apply localized modifications without re-creating the entire image.

Can I apply batch image-to-image edits to multiple product photos at once?

Yes, you can process batch image-to-image edits for up to 20 product photos simultaneously. Provide 1 to 20 HTTPS input image URLs and a prompt stating preservation first and edits last to generate consistent SKU variants.

How do I perform spatial edits on specific regions of an image?

To perform spatial edits on specific regions, use a prompt that targets localized areas like "left object only" or "upper-right corner." This allows precise modifications while locking framing via aspect ratio and resolution.

Do I need RunComfy CLI to execute Nano Banana 2 image editing workflows?

Yes, executing Nano Banana 2 image editing workflows requires invoking the RunComfy CLI. The process runs the google/nano-banana-2/edit endpoint locally and automatically downloads the edited results to an output directory.

When should I use Nano Banana 2 instead of Flux Kontext or GPT Image 2 for image editing?

Use Nano Banana 2 for identity-preserving background swaps and batch workflows. Route to GPT Image 2 for in-image text edits, or Flux Kontext for precise single-reference local edits instead of general image-to-image modifications.

Why do my localized edits fail to preserve the original subject's face and pose?

Localized edits fail to preserve subject identity if the prompt does not state preservation requirements first. Structure prompts with preservation directives before edits, and lock framing using aspect ratio and resolution settings.