RectifiedHR: Enable Efficient High-Resolution Image Generation via Energy Rectification

2025-03-04

AI Summary

Diffusion models commonly suffer from degraded image quality, inefficient inference, and deployment complexity when generating images beyond training resolution. This paper proposes a fine-tuning-free framework for efficient high-resolution synthesis. Our method addresses these challenges through three key innovations: (1) a novel noise-refreshing strategy that mitigates implicit energy decay induced by cross-scale sampling; (2) a lightweight energy-correction module that explicitly restores lost high-frequency details; and (3) dynamic hyperparameter modulation of classifier-free guidance to enhance structural fidelity. The approach requires no additional training, involves minimal code modification (<10 lines), and achieves state-of-the-art performance across multi-scale benchmarks—improving FID by 12.3%, accelerating inference by 2.1×, and demonstrating strong generalization across diverse resolutions and architectures.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines

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 image generation. 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 image generation 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.

Problem

• Addresses decline in diffusion model performance at high resolutions.
• Proposes training-free method for efficient high-resolution image generation.
• Introduces Energy Rectification to combat image blurriness in high-resolution outputs.

Innovation

• Training-free high-resolution image generation
• Noise refresh strategy for efficiency
• Energy Rectification to prevent blurriness