GeminiWatermarkTool  by allenk

AI image watermark removal utility

Created 1 month ago
807 stars

Top 43.8% on SourcePulse

GitHubView on GitHub
Project Summary

A lightweight, standalone command-line utility designed to remove visible Gemini Nano/Pro watermarks from images. It offers a fast, offline, and cross-platform solution for users needing to clean up images for slides, documents, or UI screenshots, providing deterministic restoration without generative inpainting.

How It Works

The tool employs mathematically accurate reverse alpha blending to remove watermarks. By solving the alpha blending equation (original = (watermarked - alpha_map) / (1 - α)), it reconstructs original pixels precisely, avoiding the artifacts common in generative inpainting methods. This approach ensures text remains crisp and image quality is preserved. It also automatically detects and handles both 48x48 and 96x96 watermark sizes.

Quick Start & Requirements

Pre-built, zero-dependency executables are available for Windows, Linux, macOS, and Android. The simplest usage involves dragging an image file onto the executable (Windows) or running GeminiWatermarkTool image.jpg from the command line for in-place editing. More control is available via options like -i for input and -o for output, supporting single files or batch directory processing. System requirements vary by platform, generally requiring modern OS versions.

Highlighted Details

  • Zero Dependencies: Distributed as a single, standalone executable for easy deployment.
  • Cross-Platform: Native support for Windows, Linux, macOS, and Android.
  • Deterministic Restoration: Utilizes reverse alpha blending, not AI inpainting, for pixel-perfect recovery.
  • Automatic Watermark Detection: Intelligently identifies and processes both 48x48 and 96x96 watermark sizes.
  • Batch Processing: Capable of processing entire directories of images efficiently.

Maintenance & Community

The project is authored by Allen Kuo (@allenk). While specific community channels like Discord or Slack are not detailed, the project encourages engagement via GitHub stars. No information on sponsorships, partnerships, or a public roadmap is provided.

Licensing & Compatibility

The tool is released under the permissive MIT License. This license allows for broad compatibility, including commercial use and integration into closed-source projects without significant restrictions.

Limitations & Caveats

This tool exclusively removes Gemini's visible watermarks and does not affect invisible watermarks like SynthID, which are integrated during image generation and cannot be separated. The tool is designed for Gemini's current watermark implementation (as of 2025), and variations may lead to unexpected results. Users are strongly advised to back up original images, as the tool operates on a "use at your own risk" basis, with the author disclaiming responsibility for any data loss or image corruption.

Health Check
Last Commit

1 week ago

Responsiveness

Inactive

Pull Requests (30d)
2
Issues (30d)
9
Star History
817 stars in the last 30 days

Explore Similar Projects

Starred by Robin Huang Robin Huang(Cofounder of Comfy Org), Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), and
17 more.

stablediffusion by Stability-AI

0%
42k
Latent diffusion model for high-resolution image synthesis
Created 3 years ago
Updated 6 months ago
Starred by Dan Abramov Dan Abramov(Core Contributor to React; Coauthor of Redux, Create React App), Patrick von Platen Patrick von Platen(Author of Hugging Face Diffusers; Research Engineer at Mistral), and
57 more.

stable-diffusion by CompVis

0.1%
72k
Latent text-to-image diffusion model
Created 3 years ago
Updated 1 year ago
Feedback? Help us improve.