MaXsive: High-Capacity and Robust Training-Free Generative Image Watermarking in Diffusion Models

TL;DR

MaXsive introduces a training-free, high-capacity, and robust image watermarking method for diffusion models, effectively resisting RST attacks and reducing ID collusion, verified on standard benchmarks.

Contribution

It proposes a novel watermarking technique that uses initial noise and an X-shape template to enhance robustness and capacity without training, addressing prior vulnerabilities.

Findings

MaXsive achieves high robustness against RST attacks.

The method maintains high watermark capacity.

It reduces identity collusion risk.

Abstract

The great success of the diffusion model in image synthesis led to the release of gigantic commercial models, raising the issue of copyright protection and inappropriate content generation. Training-free diffusion watermarking provides a low-cost solution for these issues. However, the prior works remain vulnerable to rotation, scaling, and translation (RST) attacks. Although some methods employ meticulously designed patterns to mitigate this issue, they often reduce watermark capacity, which can result in identity (ID) collusion. To address these problems, we propose MaXsive, a training-free diffusion model generative watermarking technique that has high capacity and robustness. MaXsive best utilizes the initial noise to watermark the diffusion model. Moreover, instead of using a meticulously repetitive ring pattern, we propose injecting the X-shape template to recover the RST distortions. This design significantly increases robustness without losing any capacity, making ID collusion less likely to happen. The effectiveness of MaXsive has been verified on two well-known watermarking benchmarks under the scenarios of verification and identification.