Spectral Amplitude Purification in Distribution Matching for Diffusion Distillation

2026-09-24

AI Summary

This study addresses the challenge in diffusion model distillation where low-frequency signal dominance impedes high-frequency detail recovery, causing directional errors to readily fall into low-frequency overfitting. To overcome this limitation, this work proposes Spectral Amplitude Purification Distribution Matching Distillation (SAP-DMD), a novel framework that introduces a spectral amplitude purification mechanism for the first time. By integrating Fourier spectral analysis with adaptive amplitude modulation, the proposed approach effectively mitigates low-frequency dominance while enhancing high-frequency reconstruction. Extensive evaluations on models such as PixArt demonstrate that SAP-DMD significantly accelerates convergence and substantially improves image generation quality under few-step sampling regimes. Consequently, this research establishes a new frequency-domain optimization paradigm for efficient diffusion distillation.

Abstract

Distribution Matching Distillation (DMD) enables high-quality diffusion sampling in only a few steps, but its optimization dynamics remain dominated by coarse, low-frequency signals, delaying the recovery of fine-grained details. We identify a pronounced concentration of spectral amplitudes at low frequencies in the DMD directional error, where dominant low-frequency components overwhelm weaker mid- and high-frequency signals. To address this issue, we propose Spectral Amplitude Purification for Distribution Matching Distillation (SAP-DMD), a plug-and-play approach that adaptively modulates the amplitude spectrum of the DMD directional field. By suppressing the dominant tail of the amplitude spectrum, SAP-DMD reduces low-frequency dominance and promotes more effective recovery of fine structures and textures. Experiments on PixArt-$α$, SD3, and SD3.5 demonstrate that SAP-DMD accelerates training convergence and improves generation quality under both 2-step and 4-step sampling.

Problem

• Diffusion Distillation
• Distribution Matching Distillation
• Spectral Amplitude
• Low-frequency Dominance
• Fine-grained Details

Innovation

• Distribution Matching Distillation
• Spectral Amplitude Purification
• Diffusion Distillation
• Frequency Domain Analysis
• Plug-and-Play