RectifiedHR: Enable Efficient High-Resolution Synthesis via Energy Rectification
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917392957833216 |
|---|---|
| author | Yang, Zhen Shen, Guibao Li, Minyang Hou, Liang Liu, Mushui Wang, Luozhou Tao, Xin Chen, Ying-Cong |
| author_facet | Yang, Zhen Shen, Guibao Li, Minyang Hou, Liang Liu, Mushui Wang, Luozhou Tao, Xin Chen, Ying-Cong |
| contents | Diffusion models have achieved remarkable progress across various visual generation tasks. However, their performance significantly declines when generating content at resolutions higher than those used during training. Although numerous methods have been proposed to enable high-resolution generation, they all suffer from inefficiency. In this paper, we propose RectifiedHR, a straightforward and efficient solution for training-free high-resolution synthesis. Specifically, we propose a noise refresh strategy that unlocks the model's training-free high-resolution synthesis capability and improves efficiency. Additionally, we are the first to observe the phenomenon of energy decay, which may cause image blurriness during the high-resolution synthesis process. To address this issue, we introduce average latent energy analysis and find that tuning the classifier-free guidance hyperparameter can significantly improve generation performance. Our method is entirely training-free and demonstrates efficient performance. Furthermore, we show that RectifiedHR is compatible with various diffusion model techniques, enabling advanced features such as image editing, customized generation, and video synthesis. Extensive comparisons with numerous baseline methods validate the superior effectiveness and efficiency of RectifiedHR. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_02537 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | RectifiedHR: Enable Efficient High-Resolution Synthesis via Energy Rectification Yang, Zhen Shen, Guibao Li, Minyang Hou, Liang Liu, Mushui Wang, Luozhou Tao, Xin Chen, Ying-Cong Computer Vision and Pattern Recognition Artificial Intelligence Diffusion models have achieved remarkable progress across various visual generation tasks. However, their performance significantly declines when generating content at resolutions higher than those used during training. Although numerous methods have been proposed to enable high-resolution generation, they all suffer from inefficiency. In this paper, we propose RectifiedHR, a straightforward and efficient solution for training-free high-resolution synthesis. Specifically, we propose a noise refresh strategy that unlocks the model's training-free high-resolution synthesis capability and improves efficiency. Additionally, we are the first to observe the phenomenon of energy decay, which may cause image blurriness during the high-resolution synthesis process. To address this issue, we introduce average latent energy analysis and find that tuning the classifier-free guidance hyperparameter can significantly improve generation performance. Our method is entirely training-free and demonstrates efficient performance. Furthermore, we show that RectifiedHR is compatible with various diffusion model techniques, enabling advanced features such as image editing, customized generation, and video synthesis. Extensive comparisons with numerous baseline methods validate the superior effectiveness and efficiency of RectifiedHR. |
| title | RectifiedHR: Enable Efficient High-Resolution Synthesis via Energy Rectification |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2503.02537 |