KITTEN: A Knowledge-Intensive Evaluation of Image Generation on Visual Entities

TL;DR

KITTEN is a benchmark designed to evaluate how well current text-to-image models can accurately generate detailed visual entities from real-world knowledge, revealing limitations in their fidelity and creativity.

Contribution

This paper introduces KITTEN, a novel benchmark for knowledge-intensive image generation, and provides a systematic evaluation of state-of-the-art models' capabilities and limitations.

Findings

Advanced models often fail to capture detailed visual features.

Retrieval-augmented models improve fidelity but over-rely on reference images.

Models struggle with generating novel configurations of entities.

Abstract

Recent advances in text-to-image generation have improved the quality of synthesized images, but evaluations mainly focus on aesthetics or alignment with text prompts. Thus, it remains unclear whether these models can accurately represent a wide variety of realistic visual entities. To bridge this gap, we propose KITTEN, a benchmark for Knowledge-InTensive image generaTion on real-world ENtities. Using KITTEN, we conduct a systematic study of the latest text-to-image models and retrieval-augmented models, focusing on their ability to generate real-world visual entities, such as landmarks and animals. Analysis using carefully designed human evaluations, automatic metrics, and MLLM evaluations show that even advanced text-to-image models fail to generate accurate visual details of entities. While retrieval-augmented models improve entity fidelity by incorporating reference images, they tend to over-rely on them and struggle to create novel configurations of the entity in creative text prompts.