Detection Limits and Statistical Separability of Tree Ring Watermarks in Rectified Flow-based Text-to-Image Generation Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Umrajkar, Ved, Singh, Aakash Kumar
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908302694154240
author Umrajkar, Ved
Singh, Aakash Kumar
author_facet Umrajkar, Ved
Singh, Aakash Kumar
contents Tree-Ring Watermarking is a significant technique for authenticating AI-generated images. However, its effectiveness in rectified flow-based models remains unexplored, particularly given the inherent challenges of these models with noise latent inversion. Through extensive experimentation, we evaluated and compared the detection and separability of watermarks between SD 2.1 and FLUX.1-dev models. By analyzing various text guidance configurations and augmentation attacks, we demonstrate how inversion limitations affect both watermark recovery and the statistical separation between watermarked and unwatermarked images. Our findings provide valuable insights into the current limitations of Tree-Ring Watermarking in the current SOTA models and highlight the critical need for improved inversion methods to achieve reliable watermark detection and separability. The official implementation, dataset release and all experimental results are available at this \href{https://github.com/dsgiitr/flux-watermarking}{\textbf{link}}.
format Preprint
id arxiv_https___arxiv_org_abs_2504_03850
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Detection Limits and Statistical Separability of Tree Ring Watermarks in Rectified Flow-based Text-to-Image Generation Models
Umrajkar, Ved
Singh, Aakash Kumar
Computer Vision and Pattern Recognition
Artificial Intelligence
Cryptography and Security
Machine Learning
Tree-Ring Watermarking is a significant technique for authenticating AI-generated images. However, its effectiveness in rectified flow-based models remains unexplored, particularly given the inherent challenges of these models with noise latent inversion. Through extensive experimentation, we evaluated and compared the detection and separability of watermarks between SD 2.1 and FLUX.1-dev models. By analyzing various text guidance configurations and augmentation attacks, we demonstrate how inversion limitations affect both watermark recovery and the statistical separation between watermarked and unwatermarked images. Our findings provide valuable insights into the current limitations of Tree-Ring Watermarking in the current SOTA models and highlight the critical need for improved inversion methods to achieve reliable watermark detection and separability. The official implementation, dataset release and all experimental results are available at this \href{https://github.com/dsgiitr/flux-watermarking}{\textbf{link}}.
title Detection Limits and Statistical Separability of Tree Ring Watermarks in Rectified Flow-based Text-to-Image Generation Models
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2504.03850