发表机构
Sony AI; Sony Group Corporation（索尼人工智能; 索尼集团公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对扩散模型曝光偏差问题，提出频谱对齐（SPA）方法，通过离线拟合参数频谱模型及推理时基于快速傅里叶变换的梯度计算引导，减少计算开销且与CFG互补，在多种架构上实现一致改进。
AI 中文摘要
扩散模型在迭代采样过程中通常会受到误差积累的影响，即曝光偏差。我们揭示了训练和推理之间系统的频率依赖性差异，可解释为频率依赖性信噪比误差。关键的是，这种不匹配的方向在模型和时间步长之间变化，这表明固定的校正规则不具有通用性。我们提出了频谱对齐（SPA）方法，这是一种轻量级的基于引导的方法，将中间预测的功率谱校准到预先计算的先验。我们的方法包括两个阶段：（1）从训练数据中离线拟合参数频谱模型；（2）在推理时通过基于快速傅里叶变换的高效梯度计算进行引导。SPA引入的计算开销最小（3-4%），并且与无分类器引导（CFG）互补。我们展示了从像素空间模型（DDPM、ADM）到潜在扩散模型（SD2.0、SDXL）和流匹配模型（SD3.5、FLUX）等各种架构的一致改进。我们的实现可在这个https URL上获取。
英文摘要
Diffusion models typically suffer from error accumulation during iterative sampling, commonly referred to as exposure bias. We reveal systematic frequency-dependent discrepancies between training and inference, which can be interpreted as frequency-dependent SNR error. Crucially, the direction of this mismatch varies across models and timesteps, indicating that fixed correction rules do not generalize. We propose Spectral Alignment (SPA), a lightweight, guidance-based method that calibrates the power spectrum of intermediate predictions to a pre-computed prior. Our approach consists of two stages: (1) offline fitting of a parametric spectrum model from training data, and (2) inference-time guidance via efficient FFT-based gradient computation. SPA introduces minimal computational overhead (3-4\%) and is complementary to Classifier-Free Guidance (CFG). We demonstrate consistent improvements across diverse architectures, from pixel-space models (DDPM, ADM) to latent diffusion models (SD2.0, SDXL) and flow-matching models (SD3.5, FLUX). Our implementation is available at https://github.com/SonyResearch/SPA.
CommentsAccepted at ECCV2026
