Research map / Image generation / Autoregressive and latent generation
RAIN: Redundancy-Aware Latent Injection for Quality-Preserving Image Watermarking
Yehan Sun, Rongrong Ni, Chuangchuang Tan, Huan Liu, Wenhao Ni, Renshuai Tao, Yao Zhao
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 redundancy-aware guidance (ablation) | RAIN · ablation, 10k iterations | image PSNR: 20.44 -> 24.20 | Improved |
| add redundancy-aware guidance with alignment (ablation) | RAIN · ablation, 10k iterations | image PSNR: 23.88 -> 26.45 | Improved |
| add distribution alignment initialization (ablation) | RAIN · ablation, 10k iterations | watermark recovery PSNR: 35.57 -> 41.58 | Improved |
| add distribution alignment initialization (ablation) | RAIN · ablation, 2k iterations | watermark recovery PSNR: 14.26 -> 35.44 | Improved |
| add distribution alignment initialization (ablation) | RAIN · ablation, 2k iterations | watermark extraction ACC: 0.00 -> 0.71 | Improved |
| add distribution alignment initialization (ablation) | RAIN · ablation | convergence iterations: 25k -> 13k | Improved |
| add redundancy guidance (ablation) | RAIN · ablation, 10k iterations | watermark recovery PSNR: 35.57 -> 35.94 | Improved |
| use both redundancy guidance and alignment (ablation) | RAIN · ablation, 10k iterations | watermark ACC: 0.70 -> 0.85 | Improved |
| use RAIN instead of fine-tuning-based baselines | RAIN · COCO2017 | FID change from original: -0.0226 vs +3.2809 (RoSteALS) | Improved |
| apply to other SD versions without retraining | RAIN · SD1-4, SD1-5, SD2-0, SD2-1 | PSNR: 39.9330 -> 40.6300 | No clear effect |
| use RAIN under crop attack | RAIN · robustness | ACC | 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.
- SleeperMark — SleeperMark: Towards Robust Watermark against Fine-Tuning Text-to-image Diffusion Models (CVPR 2025) · 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
- Black-Box Forgery Attacks on Semantic Watermarks for Diffusion Models (CVPR 2025 · similarity 0.87)
- Shallow Diffuse: Robust and Invisible Watermarking through Low-Dim Subspaces in Diffusion Models (NeurIPS 2025 · similarity 0.87)
- Image Watermarks are Removable using Controllable Regeneration from Clean Noise (ICLR 2025 · similarity 0.87)
- WMAdapter: Adding WaterMark Control to Latent Diffusion Models (ICML 2025 · similarity 0.87)
- T2SMark: Balancing Robustness and Diversity in Noise-as-Watermark for Diffusion Models (NeurIPS 2025 · similarity 0.87)
- StableGuard: Towards Unified Copyright Protection and Tamper Localization in Latent 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.
