GLOW-FM: One-Step Latent Flow Matching for Low-Light Enhancement
October 22, 2026
GLOW-FM is a one-step latent flow-matching framework designed for efficient generative low-light image enhancement. It uses a frozen Tiny-VAE and Terminal Velocity Matching to achieve high-quality image recovery in a single function evaluation, reducing the computational overhead of traditional iterative diffusion methods.
HOW THIS AFFECTS YOU
●
builderYou can deploy high-quality image enhancement at much lower latency than diffusion models.
●
researcherThis provides a method for one-step generative tasks via latent flow matching.