Research map / Image generation / Autoregressive and latent generation
SynTag: Enhancing the Geometric Robustness of Inversion-based Generative Image Watermarking
Han Fang, Kejiang Chen, Zehua Ma, Jiajun Deng, Yicong Li, Weiming Zhang, Ee-Chien Chang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add SynTag injection and correction | GauShad · Geometric distortions SD-v1.4 | TPR: 0.016 -> 0.980 | Improved |
| add SynTag injection and correction | TreeRings · Geometric distortions SD-v1.4 | TPR: 0.548 -> 0.928 | Improved |
| add SynTag injection | GauShad · Geometric distortions SD-v1.4 | bit accuracy: 0.633 -> 0.938 | Improved |
| add SynTag injection | GauShad · Non-geometric distortions SD-v1.4 | TPR: 0.963 -> 0.967 | No clear effect |
| add SynTag injection | GauShad · Image quality | FID: 25.23 -> 25.21 | No clear effect |
| add SynTag decoder with predictor instead of passive predictor only (ablation) | geometric correction · 4 geometric distortions | bit accuracy: 0.730 -> 0.938 | Improved |
| add SynTag predictor PSyn (ablation) | GauShad-SynTag · 4 geometric distortions | TPR: 0.016 -> 0.763 | Improved |
| add SynTag predictor PSyn (ablation) | GauShad-SynTag · 4 geometric distortions | bit accuracy: 0.633 -> 0.819 | Improved |
| add pixel-level dither compensation Cp (ablation) | PSyn · 4 geometric distortions | TPR: 0.763 -> 0.795 | Improved |
| add latent-level dither compensation Cl (ablation) | PSyn · 4 geometric distortions | TPR: 0.763 -> 0.930 | Improved |
| add both Cp and Cl (ablation) | PSyn · 4 geometric distortions | TPR: 0.763 -> 0.990 | Improved |
| increase purification attack strength from 0.3 to 0.7 (ablation) | GauShad-SynTag · Purification attack | TPR: 0.980 -> 0.000 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LaWa — LaWa: Using Latent Space for In-Generation Image Watermarking (ECCV 2024) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- CoSDA: Enhancing the Robustness of Inversion-based Generative Image Watermarking Framework (AAAI 2025 · similarity 0.90)
- Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advances (ICLR 2025 · similarity 0.85)
- PGID: Progressive Guided Inversion and Denoising for Robust Watermark Detection (NeurIPS 2026 · similarity 0.85)
- A Locally Tokenized Generative Model for Robust Time-Series Watermarking (NeurIPS 2026 · similarity 0.85)
- T2SMark: Balancing Robustness and Diversity in Noise-as-Watermark for Diffusion Models (NeurIPS 2025 · similarity 0.85)
- Watermarking Autoregressive Image Generation (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
