Iterative Flow Matching -- Path Correction and Gradual Refinement for Enhanced Generative Modeling

TL;DR

This paper introduces an iterative flow matching method that improves image generation quality by reducing hallucinations and can be integrated into various generative models for more robust synthesis.

Contribution

It proposes an iterative process for flow matching that enhances image generation, addressing hallucinations and improving robustness across different generative techniques.

Findings

Iterative flow matching reduces hallucinations in generated images.

The method improves robustness and performance of image synthesis.

Applicable to various generative modeling techniques.

Abstract

Generative models for image generation are now commonly used for a wide variety of applications, ranging from guided image generation for entertainment to solving inverse problems. Nonetheless, training a generator is a non-trivial feat that requires fine-tuning and can lead to so-called hallucinations, that is, the generation of images that are unrealistic. In this work, we explore image generation using flow matching. We explain and demonstrate why flow matching can generate hallucinations, and propose an iterative process to improve the generation process. Our iterative process can be integrated into virtually $\textit{any}$ generative modeling technique, thereby enhancing the performance and robustness of image synthesis systems.