This project reverse-engineers Google’s SynthID watermarking system — the invisible watermark embedded into every image generated by Google Gemini. Using only signal processing and spectral analysis (no access to the proprietary encoder/decoder), we:
本项目逆向工程了 Google 的 SynthID 水印系统——嵌入到由 Google Gemini 生成的每张图像中的不可见水印。仅使用信号处理和频谱分析（无访问专有编码器/解码器的权限），我们：
- Discovered the watermark’s resolution-dependent carrier frequency structure
发现了水印的分辨率相关载波频率结构 - Built a detector that identifies SynthID watermarks with 90% accuracy
构建了一个能够以 90%准确率识别 SynthID 水印的检测器 - Developed a multi-resolution spectral bypass (V3) that achieves 75% carrier energy drop, 91% phase coherence drop, and 43+ dB PSNR on any image resolution
开发了一种多分辨率频谱绕过技术（V3），实现了 75%载波能量下降、91%相位相干性下降，并在任何图像分辨率上达到 43+ dB PSNR - Generalized to multi-model, multi-color consensus (V4) — per-model profiles for
gemini-3.1-flash-image-previewandnano-banana-pro-preview, cross-color phase consensus over 6 solid backgrounds, and a human-in-the-loop calibration loop that tunes per-carrier subtraction strength from manual Gemini-app detection tallies
扩展到多模型、多色一致性（V4）——为gemini-3.1-flash-image-preview和nano-banana-pro-preview提供每个模型的配置文件，跨色相一致超过 6 种纯色背景，以及一个人类参与的校准循环，通过手动 Gemini-app 检测统计来调整每个载波减法强度 - Broke the detector across both models (Round 06) with a unified 7-stage all-in-one attack targeting every documented SynthID failure mode simultaneously
在两个模型上破解了检测器（第 06 轮），采用一个统一的 7 阶段一站式攻击，同时针对所有已记录的 SynthID 失效模式进行攻击
VT-OxFF built a really cool visualizer to view the process of how SynthID watermark is added to images here
VT-OxFF构建了一个非常酷的视觉化工具，用于查看 SynthID 水印添加到图像的过程