Pixel-Space Diffusion via Observation Operators

2026-08-22

AI Summary

本文针对像素空间扩散模型优化难题，提出观察算子扩散框架，通过从粗到细的监督轨迹与特征细化，提高生成质量并加速收敛。

Abstract

Pixel-space diffusion models directly model image distributions but remain difficult to optimize. Recent methods alleviate this challenge through target reparameterization, while still relying on a fixed clean-image target throughout denoising. Through empirical analysis, we identify a scale-time mismatch: image structures become predictable from coarse to fine as noise decreases, whereas existing models are forced to predict the full image even under high noise, resulting in low-SNR gradients that hinder optimization. To resolve this mismatch, we propose Observation Operator Diffusion, a unified framework that aligns both the supervision trajectory and feature refinement with the intrinsic recovery order of image structures. Specifically, we replace fixed full-image supervision along the standard flow path with a time-indexed observation trajectory that evolves from coarse structures to the full image during denoising. This trajectory is instantiated with a family of Gaussian-Lanczos operators at varying observation scales, yielding a path-consistent training objective. We further introduce GL-CoDA, a decoder that injects scale-specific Gaussian-Lanczos observations across decoding stages for coarse-to-fine feature refinement. Extensive experiments show that the proposed approach converges substantially faster while consistently improving generation quality, achieving an FID of 1.52 on ImageNet-256.

Problem

• Pixel-space diffusion
• Optimization
• Denoising
• Low-SNR gradients
• Image structures

Innovation

• Observation Operator Diffusion
• scale-time mismatch
• Gaussian-Lanczos operators
• coarse-to-fine feature refinement
• GL-CoDA