VersaT2I: Improving Text-to-Image Models with Versatile Reward

TL;DR

VersaT2I is a versatile training framework that enhances text-to-image models by incorporating multiple quality rewards, improving aesthetics, accuracy, and fidelity without manual annotations or architecture changes.

Contribution

It introduces a multi-aspect reward-based finetuning method with a gating function, enabling improved T2I performance across various quality metrics.

Findings

Outperforms baseline models on multiple quality benchmarks

Effectively combines multiple quality aspects without conflicts

Does not require manual annotations or architecture modifications

Abstract

Recent text-to-image (T2I) models have benefited from large-scale and high-quality data, demonstrating impressive performance. However, these T2I models still struggle to produce images that are aesthetically pleasing, geometrically accurate, faithful to text, and of good low-level quality. We present VersaT2I, a versatile training framework that can boost the performance with multiple rewards of any T2I model. We decompose the quality of the image into several aspects such as aesthetics, text-image alignment, geometry, low-level quality, etc. Then, for every quality aspect, we select high-quality images in this aspect generated by the model as the training set to finetune the T2I model using the Low-Rank Adaptation (LoRA). Furthermore, we introduce a gating function to combine multiple quality aspects, which can avoid conflicts between different quality aspects. Our method is easy to extend and does not require any manual annotation, reinforcement learning, or model architecture changes. Extensive experiments demonstrate that VersaT2I outperforms the baseline methods across various quality criteria.