Research map / Image generation / Super-resolution
Adversarial Score Distillation for Stable One-Step Diffusion in Real-World Image Super-Resolution
Wei Zhu, Kai Zhang, Yu Zheng, Lei Luo, Jian Yang
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 |
|---|---|---|---|
| compute penalty on GT latent instead of generated latent (ablation) | ASDSR · DrealSR | LPIPS: 0.3132 -> 0.2896 | Improved |
| compute penalty on GT latent instead of generated latent (ablation) | ASDSR · RealSR | LPIPS: 0.3187 -> 0.2805 | Improved |
| add LPIPS loss to training (ablation) | ASDSR · DrealSR | LPIPS: 0.4685 -> 0.2896 | Improved |
| add semantic consistency loss (CLIP) (ablation) | ASDSR · DrealSR | CLIPIQA: 0.6423 -> 0.6885 | Improved |
| add semantic consistency loss (CLIP) (ablation) | ASDSR · DrealSR | LPIPS: 0.2909 -> 0.2896 | No clear effect |
| replace VSD with ASD (ablation) | ASDSR (SD2.1-base) · DrealSR | CLIPIQA: 0.6644 -> 0.6885 | Improved |
| replace VSD with ASD (ablation) | ASDSR (SD-Turbo) · DrealSR | LPIPS: 0.2875 -> 0.2896 | Mixed |
| replace SDS with ASD (ablation) | ASDSR · DrealSR | LPIPS: 0.2853 -> 0.2896 | Mixed |
| raise SCL weight from 0.2 to 1.0 (ablation) | ASDSR · DrealSR | MUSIQ: 64.83 -> 62.18 | Hurt |
| use LoRA rank 4 instead of 2 (ablation) | ASDSR · DrealSR | LPIPS: 0.3367 -> 0.2896 | Improved |
| use LoRA rank 8 instead of 4 (ablation) | ASDSR · DrealSR | CLIPIQA: 0.6885 -> 0.6629 | Hurt |
| use CLIP instead of DINOv2 or MAE for semantic loss (ablation) | ASDSR · DrealSR | LPIPS: 0.2945/0.2932 -> 0.2896 | Improved |
Closest papers by meaning
- TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution (CVPR 2025 · similarity 0.92)
- Adversarial Diffusion Compression for Real-World Image Super-Resolution (CVPR 2025 · similarity 0.91)
- Unleashing the Power of One-Step Diffusion based Image Super-Resolution via a Large-Scale Diffusion Discriminator (NeurIPS 2025 · similarity 0.90)
- Noise-Started One-Step Image Super-Resolution via LR-Conditioned SplitMeanFlow and GAN Refinement (NeurIPS 2026 · similarity 0.89)
- NSARM: Next-Scale Autoregressive Modeling for Robust Real-World Image Super-Resolution (NeurIPS 2026 · similarity 0.89)
- 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.
