Connexion
z1w2r3

nanobanana-watermark-remover

by @z1w2r3 in Tools
0
0
# Install this skill:
npx skills add z1w2r3/nanobanana-watermark-remover-skill

Or install specific skill: npx add-skill https://github.com/z1w2r3/nanobanana-watermark-remover-skill

# Description

|

# SKILL.md

name: nanobanana-watermark-remover
description: |
Remove watermarks from images generated by Google's NanoBanana AI image generator.
NanoBanana adds a distinctive four-pointed star watermark (light gray/white) typically
in the bottom-right corner of generated images.

MANDATORY TRIGGERS:
- "remove watermark", "去水印", "去除水印"
- "nanobanana", "nano banana", "Gemini Generated Image"
- Images with star-shaped watermarks in corners
- AI-generated images needing watermark removal

NanoBanana Watermark Remover

Remove the four-pointed star watermark from images generated by Google's NanoBanana AI.

Watermark Characteristics

  • Shape: Four-pointed star (✦)
  • Color: Light gray/white (semi-transparent)
  • Location: Usually bottom-right corner
  • Size: Approximately 100-150 pixels

Quick Usage

# Single image (default: bottom-right corner)
python scripts/remove_watermark.py image.png

# Specify output file
python scripts/remove_watermark.py image.png -o clean.png

# Different corner locations
python scripts/remove_watermark.py image.png --corner bl  # bottom-left
python scripts/remove_watermark.py image.png --corner tr  # top-right
python scripts/remove_watermark.py image.png --corner tl  # top-left

# Batch processing
python scripts/remove_watermark.py --batch ./photos/
python scripts/remove_watermark.py --batch ./photos/ --pattern "*.jpg"

Parameters

Parameter Default Description
--corner br Watermark position: br/bl/tr/tl
--region 200 Search region size (pixels)
--threshold 100 Brightness threshold for detection
--dark false Use if watermark is dark instead of light
-o, --output auto Output filename (default: adds _no_watermark)
--batch - Process all images in directory
--pattern *.png File pattern for batch mode

How It Works

  1. Extracts the corner region where watermark is expected
  2. Detects pixels brighter than threshold (the light watermark)
  3. Creates a mask of the watermark area
  4. Uses OpenCV's inpainting algorithm to fill the masked area with surrounding pixels

Dependencies

Ensure these Python packages are installed:

pip install opencv-python-headless numpy

Troubleshooting

Watermark not fully removed?
- Try increasing --region size (e.g., --region 250)
- Adjust --threshold (lower for fainter watermarks, higher for more visible ones)

Removing too much of the image?
- Decrease --region size
- Increase --threshold

Dark watermark instead of light?
- Use --dark flag

# README.md

NanoBanana Watermark Remover Skill

English | 中文

English

An AI agent skill for removing watermarks from images generated by Google's NanoBanana AI image generator. Compatible with Claude Code, Cursor, Windsurf, and other AI coding assistants.

Features

  • 🌟 Remove four-pointed star watermarks from NanoBanana AI-generated images
  • 📍 Support for all corner positions (bottom-right, bottom-left, top-right, top-left)
  • 🎨 Handle both light and dark watermarks
  • 📦 Batch processing support
  • 🔧 Configurable detection parameters

Installation

Option 1: Project-level installation

npx skills add z1w2r3/nanobanana-watermark-remover-skill

Option 2: Global installation

npx skills add z1w2r3/nanobanana-watermark-remover-skill --global

Option 3: Manual installation

# Clone to personal skills directory
git clone https://github.com/z1w2r3/nanobanana-watermark-remover-skill.git ~/.claude/skills/nanobanana-watermark-remover

# Or clone to project skills directory
git clone https://github.com/z1w2r3/nanobanana-watermark-remover-skill.git .claude/skills/nanobanana-watermark-remover

Usage

Natural language (Claude will auto-detect):
- "Help me remove the watermark from this image"
- "去除这张图片的水印"
- "Remove nanobanana watermark"

Command line:

# Single image
python scripts/remove_watermark.py image.png

# Specify output
python scripts/remove_watermark.py image.png -o clean.png

# Batch processing
python scripts/remove_watermark.py --batch ./photos/

Parameters

Parameter Default Description
--corner br Watermark position: br/bl/tr/tl
--region 200 Search region size (pixels)
--threshold 100 Brightness threshold
--dark false Use for dark watermarks
--batch - Batch process directory
--pattern *.png File pattern for batch mode

Requirements

  • Python 3.8+
  • OpenCV (pip install opencv-python-headless)
  • NumPy (pip install numpy)

License

MIT License

中文

一个用于去除 Google NanoBanana AI 图像生成器水印的 AI 代理技能。兼容 Claude Code、Cursor、Windsurf 等 AI 编程助手。

特性

  • 🌟 去除 NanoBanana AI 生成图片的四角星水印
  • 📍 支持所有角落位置（右下、左下、右上、左上）
  • 🎨 支持浅色和深色水印
  • 📦 支持批量处理
  • 🔧 可配置检测参数

安装

方式一：项目级安装

npx skills add z1w2r3/nanobanana-watermark-remover-skill

方式二：全局安装

npx skills add z1w2r3/nanobanana-watermark-remover-skill --global

方式三：手动安装

# 克隆到个人技能目录
git clone https://github.com/z1w2r3/nanobanana-watermark-remover-skill.git ~/.claude/skills/nanobanana-watermark-remover

# 或克隆到项目技能目录
git clone https://github.com/z1w2r3/nanobanana-watermark-remover-skill.git .claude/skills/nanobanana-watermark-remover

使用方法

自然语言（Claude 会自动识别）：
- "帮我去除这张图片的水印"
- "去水印"
- "Remove nanobanana watermark"

命令行：

# 单张图片
python scripts/remove_watermark.py image.png

# 指定输出
python scripts/remove_watermark.py image.png -o clean.png

# 批量处理
python scripts/remove_watermark.py --batch ./photos/

参数说明

参数 默认值 说明
--corner br 水印位置: br(右下)/bl(左下)/tr(右上)/tl(左上)
--region 200 搜索区域大小（像素）
--threshold 100 亮度阈值
--dark false 深色水印时使用
--batch - 批量处理目录
--pattern *.png 批量处理文件模式

依赖

  • Python 3.8+
  • OpenCV (pip install opencv-python-headless)
  • NumPy (pip install numpy)

许可证

MIT License

# Related Skills

obra
executing-plans @obra

Use when you have a written implementation plan to execute in a separate session with review checkpoints

★ 32,766
obra
using-superpowers @obra

Use when starting any conversation - establishes how to find and use skills, requiring Skill...

★ 32,766

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

⚡ Amp 🚀 Antigravity 🤖 Claude Code 🦀 Clawdbot 📝 Codex ▶️ Cursor 🤖 Droid 💎 Gemini CLI 🐙 GitHub Copilot 🪿 Goose 📊 Kilo Code 🔧 Kiro CLI 💻 OpenCode 🦘 Roo Code 🌲 Trae 🏄 Windsurf

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.