Research map / Video / Video generation
MEDUSA: Motion Elimination in Diffusion Using Spectral Attack
Hongwei Yu, Daoqing Zha, Xinlong Ding, Jiawei Li, Junbao Zhuo, Qiankun Liu, Huimin Ma, Jiansheng Chen
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 |
|---|---|---|---|
| use nuclear-norm minimization instead of hard identity-matrix target (ablation) | MEDUSA on SVD · SVD (50 images) | Optical flow: 2.3325 -> 0.5980 | Improved |
| use nuclear-norm minimization instead of hard identity-matrix target (ablation) | MEDUSA on SVD · SVD (50 images) | Dynamic degree: 36.91% -> 7.23% | Improved |
| target intermediate temporal block 5 instead of block 2 (ablation) | MEDUSA on SVD · SVD (50 images) | Optical flow: 2.7708 -> 0.5980 | Improved |
| target block 5 instead of block 8 (ablation) | MEDUSA on SVD · SVD (50 images) | Dynamic degree: 15.42% -> 7.23% | Improved |
| apply JPEG compression defense to adversarial image | MEDUSA on SVD · SVD | Dynamic degree: 13.72% -> 21.93% | Hurt |
| apply bit-depth reduction defense | MEDUSA on SVD · SVD | Optical flow: 1.2246 -> 1.4172 | Hurt |
| MEDUSA vs Vid-Freeze | video attack · SVD | Optical flow: 1.9099 -> 1.2246 | Improved |
| MEDUSA vs Vid-Freeze subject consistency | video attack · LTX-Video | Subject Consistency: 86.93% vs 85.67% | Hurt |
| MEDUSA vs PhotoGuard on DynamiCrafter | video attack · DynamiCrafter | Dynamic degree: 57.21% -> 19.82% | Improved |
| MEDUSA perturbation vs random Gaussian noise | MEDUSA · image quality | SSIM: 0.5454 -> 0.8202 | Improved |
| attack induces temporal rank collapse | MEDUSA on SVD · SVD | Effective rank: 7.74 -> 2.96 | Improved |
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.
- I2VGuard — I2VGuard: Safeguarding Images against Misuse in Diffusion-based Image-to-Video Models (CVPR 2025) · better on 6 of 6 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
- I2VGuard: Safeguarding Images against Misuse in Diffusion-based Image-to-Video Models (CVPR 2025 · similarity 0.88)
- Identifying and Solving Conditional Image Leakage in Image-to-Video Diffusion Model (NeurIPS 2024 · similarity 0.87)
- Emergent Temporal Correspondences from Video Diffusion Transformers (NeurIPS 2025 · similarity 0.87)
- Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition (ICLR 2024 · similarity 0.87)
- FreeLong: Training-Free Long Video Generation with SpectralBlend Temporal Attention (NeurIPS 2024 · similarity 0.87)
- TRIP: Temporal Residual Learning with Image Noise Prior for Image-to-Video Diffusion Models (CVPR 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.
