G-Refine: A General Quality Refiner for Text-to-Image Generation

TL;DR

G-Refine is a versatile image quality refiner that enhances AI-generated images by identifying perception and alignment issues, significantly improving quality metrics and facilitating broader adoption of text-to-image models.

Contribution

We introduce G-Refine, a general quality refiner with modules for perception, alignment, and targeted enhancement, improving AI-generated image quality without degrading high-quality images.

Findings

Outperforms alternative methods on 10+ quality metrics

Effective across 4 different databases

Enhances practical application of T2I models

Abstract

With the evolution of Text-to-Image (T2I) models, the quality defects of AI-Generated Images (AIGIs) pose a significant barrier to their widespread adoption. In terms of both perception and alignment, existing models cannot always guarantee high-quality results. To mitigate this limitation, we introduce G-Refine, a general image quality refiner designed to enhance low-quality images without compromising the integrity of high-quality ones. The model is composed of three interconnected modules: a perception quality indicator, an alignment quality indicator, and a general quality enhancement module. Based on the mechanisms of the Human Visual System (HVS) and syntax trees, the first two indicators can respectively identify the perception and alignment deficiencies, and the last module can apply targeted quality enhancement accordingly. Extensive experimentation reveals that when compared to alternative optimization methods, AIGIs after G-Refine outperform in 10+ quality metrics across 4 databases. This improvement significantly contributes to the practical application of contemporary T2I models, paving the way for their broader adoption. The code will be released on https://github.com/Q-Future/Q-Refine.