Research map / 3D vision / NeRF and radiance fields
Protecting NeRFs' Copyright via Plug-And-Play Watermarking Base Model
Qi Song, Ziyuan Luo, Ka Chun Cheung, Simon See, Renjie Wan
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 |
|---|---|---|---|
| replace local-only rendering with progressive global rendering for watermark distillation (ablation) | NeRFProtector · Blender-test | bit accuracy: 45.99 -> 92.69 | Improved |
| use multi-scale progressive rendering instead of single-scale global rendering (ablation) | NeRFProtector · Blender-test | bit accuracy / PSNR: bit acc 87.27 -> 92.69; PSNR 29.57 -> 29.26 | Mixed |
| use a different extractor (MBRS) as the base model (ablation) | NeRFProtector · Blender-test | bit accuracy: 91.96 -> 89.13 | No clear effect |
| apply the watermarking base model to TensorRF and Plenoxels instead of Instant-NGP (ablation) | NeRFProtector · NeRF variants | bit accuracy: 91.96 (Instant-NGP), 89.35 (TensorRF), 97.92 (Plenoxels) | No clear effect |
| pre-watermark training images with HiDDeN or MBRS and then train NeRF | NeRF · Blender-test | bit accuracy: 50.29 (HiDDeN), 50.53 (MBRS) vs 92.69 (NeRFProtector) | Hurt |
| use NeRFProtector instead of CopyRNeRF | Blender-test · Blender-test | bit accuracy and PSNR: bit acc 62.15 -> 92.69; PSNR 25.50 -> 29.26 | Improved |
| embed watermark via base-model distillation instead of CopyRNeRF joint training | training cost · Blender-test | watermark training time: 30h -> 50m | Improved |
| apply JPEG compression or resizing to rendered views | NeRFProtector · Blender-test | bit accuracy: 92.69 -> 78.62 (JPEG) | Hurt |
| PGD white-box attack using the leaked base model | NeRFProtector · robustness | message removal | 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.
- CopyRNeRF — CopyRNeRF: Protecting the CopyRight of Neural Radiance Fields (ICCV 2023) · better on 4 of 5 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
- WateRF: Robust Watermarks in Radiance Fields for Protection of Copyrights (CVPR 2024 · similarity 0.93)
- DreaMark: Rooting Watermark in Score Distillation Sampling Generated Neural Radiance Fields (AAAI 2025 · similarity 0.90)
- Aleth-NeRF: Illumination Adaptive NeRF with Concealing Field Assumption (AAAI 2024 · similarity 0.88)
- NeRF Analogies: Example-Based Visual Attribute Transfer for NeRFs (CVPR 2024 · similarity 0.87)
- CodecNeRF: Toward Fast Encoding and Decoding, Compact, and High-quality Novel-view Synthesis (AAAI 2025 · similarity 0.87)
- A General Implicit Framework for Fast NeRF Composition and Rendering (AAAI 2024 · 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.
