Consistency Regularised Gradient Flows for Inverse Problems

arXiv:2605.07907 · stat.ML, cs.CV, cs.LG · Submitted 2026-05-08 · Read on arXiv

stat.ML, cs.CV, cs.LG

Submitted: 2026-05-08

Updated: 2026-11-09

Comments: Accepted to NeurIPS 2026

Code: https://github.com/LeviBorodenko/motionblur

License: http://creativecommons.org/licenses/by/4.0/

The gist: Vision-Language Latent Diffusion Models (LDMs) (Rombach et al., 2022) provide powerful generative priors for inverse problems.

Terminology

Abstract

Vision-Language Latent Diffusion Models (LDMs) (Rombach et al., 2022) provide powerful generative priors for inverse problems. However, existing LDM-based inverse solvers typically require a large number of neural function evaluations (NFEs) and backpropagation through large pretrained components, leading to substantial computational costs and, in some cases, degraded reconstruction quality. We propose a unified Euclidean-Wasserstein-2 gradient-flow framework that jointly performs posterior sampling and prompt optimization in the latent space through a single flow that aligns the prior and posterior with the observed data. Combined with few-step latent text-to-image models, this formulation enables low-NFE inference without backpropagation through autoencoders. Experiments across several canonical imaging inverse problems show that our method achieves state-of-the-art performance with significantly reduced computational cost.

Sources

Related papers