Detection Limits and Statistical Separability of Tree Ring Watermarks in Rectified Flow-based Text-to-Image Generation Models
Track: tiny / short paper (3-5 pages)
Keywords: Watermarking, Diffusion, Flow-Matching, Rectified Flow, Text-to-Image Generation
TL;DR: Tree Ring Watermarks are harder to detect in modern rectified flow-based models compared to traditional diffusion models, especially under image attacks.
Abstract: 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.
Presenter: ~Ved_Umrajkar1
Format: Maybe: the presenting author will attend in person, contingent on other factors that still need to be determined (e.g., visa, funding).
Funding: Yes, the presenting author of this submission falls under ICLR’s funding aims, and funding would significantly impact their ability to attend the workshop in person.
Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.
Submission Number: 33
Loading