PhaseMark: A Post-hoc, Optimization-Free Watermarking of AI-generated Images in the Latent Frequency Domain

TL;DR

PhaseMark is a fast, post-hoc watermarking method for AI-generated images that operates in the latent frequency domain, offering high resilience against attacks without compromising image quality.

Contribution

It introduces a novel, optimization-free approach that modulates phase in the latent space, significantly improving speed and robustness over existing methods.

Findings

PhaseMark is thousands of times faster than optimization-based watermarking.

It achieves state-of-the-art resilience against severe attacks.

It maintains high image quality while providing robust watermarking.

Abstract

The proliferation of hyper-realistic images from Latent Diffusion Models (LDMs) demands robust watermarking, yet existing post-hoc methods are prohibitively slow due to iterative optimization or inversion processes. We introduce PhaseMark, a single-shot, optimization-free framework that directly modulates the phase in the VAE latent frequency domain. This approach makes PhaseMark thousands of times faster than optimization-based techniques while achieving state-of-the-art resilience against severe attacks, including regeneration, without degrading image quality. We analyze four modulation variants, revealing a clear performance-quality trade-off. PhaseMark demonstrates a new paradigm where efficient, resilient watermarking is achieved by exploiting intrinsic latent properties.