How I Built a Free, 100% Client-Side Gemini & Veo Video Watermark Remover Using Inverse Alpha Math

A developer has built a free, client-side tool that removes Google Gemini and Veo watermarks using inverse alpha math, avoiding the blurring caused by generative AI inpainting. The tool reconstructs the original pixels beneath the semi-transparent logo, restoring the image with bit-for-bit clarity.

If you've generated images with Google Gemini Imagen 3 or videos with Google Veo 3 , you've probably noticed the visible semi-transparent 4-pointed sparkle logo in the bottom-right corner. Most online watermark removers use generative AI inpainting. The problem? Generative AI guesses pixels and blurs the background , creating smudgy, degraded textures and lost details. I decided to solve this differently: pure client-side pixel reconstruction with zero AI blur. Traditional AI tools attempt to "paint over" the watermark by guessing what was behind it. Because Google's watermark is semi-transparent, the original background pixels are actually preserved beneath the logo overlay. Instead of hallucinating new pixels, our engine cleanly subtracts the transparent watermark layer. Result : 100% of the authentic original picture is restored with crisp, bit-for-bit clarity and zero smudging. You can test the tool directly in your browser: 👉 Live Web App : https://geminiremove.com https://geminiremove.com I’d love to hear your feedback on the UX, performance, and features you'd like to see next