SOURCE-LINKED INTELLIGENCE

Balancing Frequencies and Pixels in Flow Matching

arXiv · AI, language, vision and robotics · article · Sep 2, 2026 · UTC

Natural images follow a $1/f^2$ spectral distribution: most signal energy lies in the low spatial frequencies, while the perceptually important structures such as textures and edges occupy sparse high-frequency bands. Pixel-space reconstruction objectives, however, treat all spatial errors uniformly, causing low frequencies to dominate the optimization signal and delaying the learning of fine-scale details. In this work, we identify this objective-level spectral imbalance as a key inefficiency in training pixel-space flow models. To address it, we propose a Focal Log-Frequency Loss (f-loss), a

Read original source ↗ Open in workspace

recordType

paper

region

Global