Jul 2026
Spectral Prior for Reducing Exposure Bias in Diffusion Models
Spectral Alignment is proposed, a lightweight, guidance-based method that calibrates the power spectrum of intermediate predictions to a pre-computed prior and is complementary to Classifier-Free Guidance (CFG).
Yuya Kobayashi, Masato Ishii, Yuhta Takida et al.
· arXiv.org · 0 citations