RectifiedHR: Enable Efficient High-Resolution Image Generation via Energy Rectification
Zhen Yang, Guibao Shen, Liang Hou, Mushui Liu, Luozhou Wang, Xin Tao, Pengfei Wan, Di Zhang, Ying-Cong Chen
2025-03-05
Summary
This paper talks about RectifiedHR, a new method for creating high-quality, high-resolution images using AI without needing additional training
What's the problem?
Current AI models for generating images struggle when trying to create pictures at higher resolutions than they were originally trained on. Existing solutions to this problem are either slow or too complicated
What's the solution?
The researchers developed RectifiedHR, which uses two main strategies. First, they introduced 'noise refresh,' a simple way to help the AI create higher-resolution images more efficiently. Second, they came up with 'Energy Rectification' to fix a problem where images become blurry during the creation process. This method adjusts certain settings in the AI to improve the quality of the generated images
Why it matters?
This matters because it allows AI to create better, higher-resolution images without needing expensive retraining or complex processes. It could make it easier and faster to generate high-quality images for various applications, from art and design to scientific visualization. The simplicity of the method also means it could be widely adopted, potentially improving image generation across many different AI systems
Abstract
Diffusion models have achieved remarkable advances in various image generation tasks. However, their performance notably declines when generating images at resolutions higher than those used during the training period. Despite the existence of numerous methods for producing high-resolution images, they either suffer from inefficiency or are hindered by complex operations. In this paper, we propose RectifiedHR, an efficient and straightforward solution for training-free high-resolution <PRE_TAG>image generation</POST_TAG>. Specifically, we introduce the noise refresh strategy, which theoretically only requires a few lines of code to unlock the model's high-resolution generation ability and improve efficiency. Additionally, we first observe the phenomenon of energy decay that may cause image blurriness during the high-resolution <PRE_TAG>image generation</POST_TAG> process. To address this issue, we propose an Energy Rectification strategy, where modifying the hyperparameters of the classifier-free guidance effectively improves the generation performance. Our method is entirely training-free and boasts a simple implementation logic. Through extensive comparisons with numerous baseline methods, our RectifiedHR demonstrates superior effectiveness and efficiency.