FiDeSR：高保真且细节保持的单步扩散超分辨率
FiDeSR: High-Fidelity and Detail-Preserving One-Step Diffusion Super-Resolution
单步扩散超分中引入潜变量残差修正与高低频自适应增强器，可在不重训练的前提下调节重建保真度与细节，为少步扩散复原提供可复用的推理期增强思路。
原文依据 FiDeSR incorporates a latent residual refinement, which corrects prediction errors in the diffusion noise
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基于扩散的方法近来推动了真实世界图像超分辨率的显著进展。然而，现有方法仍难以同时保持精细细节并确保高保真重建，往往导致次优的视觉质量。本文提出 FiDeSR，一种高保真且细节保持的单步扩散超分辨率框架。在训练阶段，我们引入细节感知加权策略，自适应地强调模型预测误差较高的区域。在推理阶段，低频与高频自适应增强器进一步精化重建结果，且无需重新训练模型，从而实现灵活的增强控制。为进一步提升重建精度，FiDeSR 引入潜变量残差精化，用于修正扩散噪声中的预测误差并增强精细细节的恢复。与现有基于扩散的方法相比，FiDeSR 取得了更优的真实世界超分辨率性能，输出兼具高感知质量与忠实的内容还原。源代码将发布于：https://github.com/Ar0Kim/FiDeSR。
英文原文摘要 1160 字符
Diffusion-based approaches have recently driven remarkable progress in real-world image super-resolution (SR). However, existing methods still struggle to simultaneously preserve fine details and ensure high-fidelity reconstruction, often resulting in suboptimal visual quality. In this paper, we propose FiDeSR, a high-fidelity and detail-preserving one-step diffusion super-resolution framework. During training, we introduce a detail-aware weighting strategy that adaptively emphasizes regions where the model exhibits higher prediction errors. During inference, low- and high-frequency adaptive enhancers further refine the reconstruction without requiring model retraining, enabling flexible enhancement control. To further improve the reconstruction accuracy, FiDeSR incorporates a latent residual refinement, which corrects prediction errors in the diffusion noise and enhances fine detail recovery. FiDeSR achieves superior real-world SR performance compared to existing diffusion-based methods, producing outputs with both high perceptual quality and faithful content restoration. The source code will be released at: https://github.com/Ar0Kim/FiDeSR.