Research map / Image generation / Autoregressive and latent generation
On the Effectiveness of Visible Watermarks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use multi-image matting and reconstruction instead of matting decomposition with ground-truth alpha | Watermark removal · Copyrights-fixed | PSNR: 15.66 -> 36.2 | Helped |
| Use multi-image matting and reconstruction instead of direct subtraction of the initial matted watermark | Watermark removal · CVPR2017-fixed | DSSIM: 0.085 -> 0.07 | Helped |
| Randomly translate the watermark across images | Watermark removal attack · CVPR17 | PSNR: 32.73 -> 32.80 | No clear effect |
| Vary watermark opacity by up to 10/255 per image | Watermark removal attack · CVPR17 | Detection rate: 98.6 -> 98.6 | No clear effect |
| Vary watermark opacity by up to 20/255 per image | Watermark removal attack · Copyright | PSNR: 36.2 -> 33.17 | Hurt |
| Apply spatial deformation of up to 0.5 pixels per image | Watermark removal attack · Copyright | PSNR: 36.2 -> 33.20 | Hurt |
| Apply spatial deformation of up to 1 pixel per image | Watermark removal attack · Copyright | DSSIM: 0.038 -> 0.085 | Hurt |
| Add the matted-watermark fidelity term (ablation) | Multi-image matting · Synthetic consistent watermark examples | Reconstruction quality | Helped |
| Estimate a per-image blend factor when opacity varies (ablation) | Generalized multi-image matting · Synthetic opacity-varying collections | Reconstruction quality | Helped |
| Estimate alpha-aware optical flow for spatially perturbed watermarks (ablation) | Generalized multi-image matting · Synthetic spatially perturbed collections | Reconstruction quality | Helped |
Closest papers by meaning
- Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advances (ICLR 2025 · similarity 0.87)
- On the Coexistence and Ensembling of Watermarks (NeurIPS 2025 · similarity 0.87)
- Transferable Black-Box One-Shot Forging of Watermarks via Image Preference Models (NeurIPS 2025 · similarity 0.87)
- WMarkGPT: Watermarked Image Understanding via Multimodal Large Language Models (ICML 2025 · similarity 0.87)
- Can Simple Averaging Defeat Modern Watermarks? (NeurIPS 2024 · similarity 0.87)
- WMCopier: Forging Invisible Watermarks on Arbitrary Images (NeurIPS 2025 · similarity 0.87)
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.