SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917226186014720 |
|---|---|
| author | Wang, Jianyi Lin, Shanchuan Lin, Zhijie Ren, Yuxi Wei, Meng Yue, Zongsheng Zhou, Shangchen Chen, Hao Zhao, Yang Yang, Ceyuan Xiao, Xuefeng Loy, Chen Change Jiang, Lu |
| author_facet | Wang, Jianyi Lin, Shanchuan Lin, Zhijie Ren, Yuxi Wei, Meng Yue, Zongsheng Zhou, Shangchen Chen, Hao Zhao, Yang Yang, Ceyuan Xiao, Xuefeng Loy, Chen Change Jiang, Lu |
| contents | Recent advances in diffusion-based video restoration (VR) demonstrate significant improvement in visual quality, yet yield a prohibitive computational cost during inference. While several distillation-based approaches have exhibited the potential of one-step image restoration, extending existing approaches to VR remains challenging and underexplored, particularly when dealing with high-resolution video in real-world settings. In this work, we propose a one-step diffusion-based VR model, termed as SeedVR2, which performs adversarial VR training against real data. To handle the challenging high-resolution VR within a single step, we introduce several enhancements to both model architecture and training procedures. Specifically, an adaptive window attention mechanism is proposed, where the window size is dynamically adjusted to fit the output resolutions, avoiding window inconsistency observed under high-resolution VR using window attention with a predefined window size. To stabilize and improve the adversarial post-training towards VR, we further verify the effectiveness of a series of losses, including a proposed feature matching loss without significantly sacrificing training efficiency. Extensive experiments show that SeedVR2 can achieve comparable or even better performance compared with existing VR approaches in a single step. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_05301 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training Wang, Jianyi Lin, Shanchuan Lin, Zhijie Ren, Yuxi Wei, Meng Yue, Zongsheng Zhou, Shangchen Chen, Hao Zhao, Yang Yang, Ceyuan Xiao, Xuefeng Loy, Chen Change Jiang, Lu Computer Vision and Pattern Recognition Recent advances in diffusion-based video restoration (VR) demonstrate significant improvement in visual quality, yet yield a prohibitive computational cost during inference. While several distillation-based approaches have exhibited the potential of one-step image restoration, extending existing approaches to VR remains challenging and underexplored, particularly when dealing with high-resolution video in real-world settings. In this work, we propose a one-step diffusion-based VR model, termed as SeedVR2, which performs adversarial VR training against real data. To handle the challenging high-resolution VR within a single step, we introduce several enhancements to both model architecture and training procedures. Specifically, an adaptive window attention mechanism is proposed, where the window size is dynamically adjusted to fit the output resolutions, avoiding window inconsistency observed under high-resolution VR using window attention with a predefined window size. To stabilize and improve the adversarial post-training towards VR, we further verify the effectiveness of a series of losses, including a proposed feature matching loss without significantly sacrificing training efficiency. Extensive experiments show that SeedVR2 can achieve comparable or even better performance compared with existing VR approaches in a single step. |
| title | SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.05301 |