Spectral and Trajectory Regularization for Diffusion Transformer Super-Resolution

TL;DR

This paper introduces StrSR, a novel one-step adversarial distillation framework with spectral and trajectory regularization, significantly improving real-world image super-resolution performance of diffusion transformers by addressing trajectory mismatch and periodic artifacts.

Contribution

The paper presents a new distillation framework with spectral and trajectory regularization to enhance diffusion transformer super-resolution, overcoming fundamental challenges of trajectory mismatch and spectral leakage.

Findings

Achieves state-of-the-art results in Real-ISR

Effectively suppresses periodic artifacts

Improves trajectory alignment in distillation process

Abstract

Diffusion transformer (DiT) architectures show great potential for real-world image super-resolution (Real-ISR). However, their computationally expensive iterative sampling necessitates one-step distillation. Existing one-step distillation methods struggle with Real-ISR on DiT. They suffer from fundamental trajectory mismatch and generate severe grid-like periodic artifacts. To tackle these challenges, we propose StrSR, a novel one-step adversarial distillation framework featuring spectral and trajectory regularization. Specifically, we propose an asymmetric discriminative distillation architecture to bridge the trajectory gap. Additionally, we design a frequency distribution matching strategy to effectively suppress DiT-specific periodic artifacts caused by high-frequency spectral leakage. Extensive experiments demonstrate that StrSR achieves state-of-the-art performance in Real-ISR, across both quantitative metrics and visual perception. The code and models will be released at https://github.com/jkwang28/StrSR .