gemini-watermark-remover

Gemini Watermark Remover is a web-based utility that removes visible watermarks from AI-generated content, specifically targeting Google Gemini, Google Flow, Gemini Omni, and Veo 3 outputs. Unlike typical AI inpainting tools that blur or hallucinate missing pixels, this tool uses a mathematical alpha unblending technique. Since the watermarks are applied with transparent alpha blending, the original pixels remain beneath the watermark. The tool subtracts the watermark mask pixel-by-pixel using the formula: Original = (Watermarked - (Logo × alpha)) / (1 - alpha), restoring the image or video to its original state with zero quality loss. Key features include: - **100% Private & Client-Side:** All processing happens locally in the browser; files never leave the device. - **Image Support:** Clean PNG, JPG, WebP images with high-resolution lossless PNG export. - **Video Support:** Process Gemini Omni, Google Flow, and Veo 3 videos frame-by-frame, export as MP4 while preserving original audio tracks. - **Live Tuner & Presets:** Quick-select presets for Nano Banana, Gemini Omni, Flow, and Veo, plus real-time sliders and dual zoomed comparison views. - **Zero Quality Loss:** Exact pixel restoration without blurry inpainting. - **Fully Responsive UI:** Works on desktop, tablet, and mobile browsers. - **Unlimited & Free:** No signups, subscriptions, or secondary watermarks. Tech stack: Pure HTML5, vanilla CSS, JavaScript (ES6+), HTML5 Canvas API, WebCodecs API, and MediaBunny for high-performance in-browser video unblending and AVC/H.264 muxing. Hosted on GitHub Pages. Local development is straightforward: clone the repo and run any static server (e.g., `python3 -m http.server 8000`). Contributions are welcome via GitHub issues and pull requests. The project is MIT licensed.