AI 中文总结
针对像素空间扩散模型的尺度-时间不匹配问题，提出观测算子扩散框架，采用时变观测轨迹与GL-CoDA解码器，在ImageNet-256上FID达1.52，收敛更快且生成质量提升。
AI 中文摘要
像素空间扩散模型直接对图像分布进行建模，但优化难度较大。现有方法通过目标重参数化缓解该挑战，但在去噪过程中仍依赖固定的干净图像目标。通过实证分析，我们发现了尺度-时间不匹配问题：随着噪声降低，图像结构从粗到细变得可预测，而现有模型即便在高噪声下也需预测完整图像，导致低信噪比梯度阻碍优化。为解决该不匹配问题，我们提出Observation Operator Diffusion（观测算子扩散），这是一个统一框架，将监督轨迹和特征细化与图像结构的固有恢复顺序对齐。具体而言，我们将标准流路径上的固定全图监督替换为时间索引的观测轨迹，该轨迹在去噪过程中从粗结构演变为全图。该轨迹由一系列不同观测尺度的Gaussian-Lanczos算子实例化，生成路径一致的训练目标。我们进一步引入GL-CoDA，这是一个解码器，在解码阶段注入特定尺度的Gaussian-Lanczos观测以实现从粗到细的特征细化。大量实验表明，所提方法收敛速度显著加快，同时持续提升生成质量，在ImageNet-256上取得了1.52的FID。
英文摘要
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.
