Research map / Image generation / Autoregressive and latent generation
RAVEN: Erasing Invisible Watermarks via Novel View Synthesis
Fahad Shamshad, Nils Lukas, Karthik Nandakumar
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 |
|---|---|---|---|
| remove watermarks by latent viewpoint shift (RAVEN) instead of UnMarker | RAVEN · MS-COCO semantic watermarks | TPR@1%FPR: 0.078 -> 0.026 | Improved |
| RAVEN vs CtrlGen+ on bit accuracy | RAVEN · MS-COCO post-hoc watermarks | bit accuracy: 0.640 -> 0.533 | Improved |
| RAVEN vs CtrlGen+ on FID for SD-Prompts | RAVEN · SD-Prompts | FID: 48.95 (CtrlGen+) vs 49.47 (RAVEN) | No clear effect |
| RAVEN vs regeneration attacks on MS-COCO quality | RAVEN · MS-COCO | FID: 42.36 (Regen) -> 40.18 | Improved |
| RAVEN vs Regen on SSIM under translation | RAVEN · DiffusionDB overlap region | SSIM: 0.74 -> 0.72 | Hurt |
| increase strength parameter s (ablation) | RAVEN · Strength sweep | FID: 62.68 (s=0.10) -> 85.10 (s=0.45) | Hurt |
| remove view-guided correspondence attention (ablation) | RAVEN · qualitative (Figure 4) | structural fidelity: no numbers given | Improved |
| add color and contrast transfer in CIELAB space (ablation) | RAVEN · Tree-Ring, HSTR, HSQR, RingID | FID: no numbers given (Figure 5) | Improved |
| swap Stable Diffusion backbone v2.1 to v1.5 (ablation) | RAVEN · MS-COCO | TPR@1%FPR: 0.026 -> 0.029 | No clear effect |
Closest papers by meaning
- Invisible Image Watermarks Are Provably Removable Using Generative AI (NeurIPS 2024 · similarity 0.89)
- Black-Box Forgery Attacks on Semantic Watermarks for Diffusion Models (CVPR 2025 · similarity 0.88)
- WMCopier: Forging Invisible Watermarks on Arbitrary Images (NeurIPS 2025 · similarity 0.88)
- Removing Interference and Recovering Content Imaginatively for Visible Watermark Removal (AAAI 2024 · similarity 0.87)
- ROBIN: Robust and Invisible Watermarks for Diffusion Models with Adversarial Optimization (NeurIPS 2024 · similarity 0.87)
- Shallow Diffuse: Robust and Invisible Watermarking through Low-Dim Subspaces in Diffusion Models (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.
