Research map / Video / Video segmentation and tracking
DiffVSR: Revealing an Effective Recipe for Taming Robust Video Super-Resolution Against Complex Degradations
Xiaohui Li, Yihao Liu, Shuo Cao, Ziyan Chen, Shaobin Zhuang, Xiangyu Chen, Yinan He, Yi Wang, Yu Qiao
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 progressive learning strategy (3 stages) instead of direct training (ablation) | DiffVSR · YouHQ40 | PSNR: 23.299 -> 24.172 | Improved |
| use progressive learning strategy instead of direct training (ablation) | DiffVSR · YouHQ40 | E*warp: 2.582 -> 1.707 | Improved |
| add stage 3 high-quality refinement to 2-stage strategy (ablation) | DiffVSR · YouHQ40 | MUSIQ: 68.449 -> 68.982 | Improved |
| use 2-stage instead of direct training (ablation) | DiffVSR · YouHQ40 | CLIP-IQA: 0.8432 -> 0.8383 | Hurt |
| add PLS (ablation) | DiffVSR with architecture · YouHQ10 | E*warp: 2.834 -> 2.056 | Improved |
| add PLS (ablation) | DiffVSR with architecture · YouHQ10 | PSNR: 23.570 -> 23.939 | Improved |
| add ILT (ablation) | DiffVSR · YouHQ10 | E*warp: 2.056 -> 1.707 | Improved |
| add ILT (ablation) | DiffVSR · YouHQ10 | MUSIQ: 69.768 -> 68.982 | Hurt |
| add architectural improvements (TE-3DVAE, MSTA) (ablation) | DiffVSR with PLS and ILT · YouHQ10 | E*warp: 2.290 -> 1.707 | Improved |
| add architectural improvements (TE-3DVAE, MSTA) (ablation) | DiffVSR with PLS and ILT · YouHQ10 | MUSIQ: 68.597 -> 68.982 | Improved |
| apply DiffVSR vs RealBasicVSR under complex degradation | DiffVSR · UDM10 (complex deg.) | MUSIQ: 60.623 -> 66.431 | Improved |
| apply DiffVSR vs RealBasicVSR under simple degradation | DiffVSR · UDM10 | PSNR: 28.526 -> 26.966 | Hurt |
Closest papers by meaning
- Motion-Guided Latent Diffusion for Temporally Consistent Real-world Video Super-resolution (ECCV 2024 · similarity 0.92)
- Unleashing the Power of One-Step Diffusion based Image Super-Resolution via a Large-Scale Diffusion Discriminator (NeurIPS 2025 · similarity 0.91)
- One-Step Diffusion for Detail-Rich and Temporally Consistent Video Super-Resolution (NeurIPS 2025 · similarity 0.90)
- DOVE: Efficient One-Step Diffusion Model for Real-World Video Super-Resolution (NeurIPS 2025 · similarity 0.90)
- PatchVSR: Breaking Video Diffusion Resolution Limits with Patch-wise Video Super-Resolution (CVPR 2025 · similarity 0.90)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.89)
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.
