Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2512.13592 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915921057021952 |
|---|---|
| author | Wang, Fu-Yun Zhou, Hao Yuan, Liangzhe Woo, Sanghyun Gong, Boqing Han, Bohyung Yang, Ming-Hsuan Zhang, Han Zhu, Yukun Liu, Ting Zhao, Long |
| author_facet | Wang, Fu-Yun Zhou, Hao Yuan, Liangzhe Woo, Sanghyun Gong, Boqing Han, Bohyung Yang, Ming-Hsuan Zhang, Han Zhu, Yukun Liu, Ting Zhao, Long |
| contents | The slow inference process of image diffusion models significantly degrades interactive user experiences. To address this, we introduce Diffusion Preview, a novel paradigm employing rapid, low-step sampling to generate preliminary outputs for user evaluation, deferring full-step refinement until the preview is deemed satisfactory. Existing acceleration methods, including training-free solvers and post-training distillation, struggle to deliver high-quality previews or ensure consistency between previews and final outputs. We propose ConsistencySolver derived from general linear multistep methods, a lightweight, trainable high-order solver optimized via Reinforcement Learning, that enhances preview quality and consistency. Experimental results demonstrate that ConsistencySolver significantly improves generation quality and consistency in low-step scenarios, making it ideal for efficient preview-and-refine workflows. Notably, it achieves FID scores on-par with Multistep DPM-Solver using 47% fewer steps, while outperforming distillation baselines. Furthermore, user studies indicate our approach reduces overall user interaction time by nearly 50% while maintaining generation quality. Code is available at https://github.com/G-U-N/consolver. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_13592 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Image Diffusion Preview with Consistency Solver Wang, Fu-Yun Zhou, Hao Yuan, Liangzhe Woo, Sanghyun Gong, Boqing Han, Bohyung Yang, Ming-Hsuan Zhang, Han Zhu, Yukun Liu, Ting Zhao, Long Machine Learning Computer Vision and Pattern Recognition The slow inference process of image diffusion models significantly degrades interactive user experiences. To address this, we introduce Diffusion Preview, a novel paradigm employing rapid, low-step sampling to generate preliminary outputs for user evaluation, deferring full-step refinement until the preview is deemed satisfactory. Existing acceleration methods, including training-free solvers and post-training distillation, struggle to deliver high-quality previews or ensure consistency between previews and final outputs. We propose ConsistencySolver derived from general linear multistep methods, a lightweight, trainable high-order solver optimized via Reinforcement Learning, that enhances preview quality and consistency. Experimental results demonstrate that ConsistencySolver significantly improves generation quality and consistency in low-step scenarios, making it ideal for efficient preview-and-refine workflows. Notably, it achieves FID scores on-par with Multistep DPM-Solver using 47% fewer steps, while outperforming distillation baselines. Furthermore, user studies indicate our approach reduces overall user interaction time by nearly 50% while maintaining generation quality. Code is available at https://github.com/G-U-N/consolver. |
| title | Image Diffusion Preview with Consistency Solver |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.13592 |