CREval: An Automated Interpretable Evaluation for Creative Image Manipulation under Complex Instructions
This addresses the problem of evaluating creative image editing models for researchers and developers, though it is incremental as it builds on existing evaluation methods.
The authors tackled the lack of systematic evaluation for creative image manipulation under complex instructions by proposing CREval, an automated QA-based pipeline, and CREval-Bench, a benchmark with over 800 samples and 13K queries, finding that all models struggle with such edits despite closed-source ones generally outperforming open-source ones.
Instruction-based multimodal image manipulation has recently made rapid progress. However, existing evaluation methods lack a systematic and human-aligned framework for assessing model performance on complex and creative editing tasks. To address this gap, we propose CREval, a fully automated question-answer (QA)-based evaluation pipeline that overcomes the incompleteness and poor interpretability of opaque Multimodal Large Language Models (MLLMs) scoring. Simultaneously, we introduce CREval-Bench, a comprehensive benchmark specifically designed for creative image manipulation under complex instructions. CREval-Bench covers three categories and nine creative dimensions, comprising over 800 editing samples and 13K evaluation queries. Leveraging this pipeline and benchmark, we systematically evaluate a diverse set of state-of-the-art open and closed-source models. The results reveal that while closed-source models generally outperform open-source ones on complex and creative tasks, all models still struggle to complete such edits effectively. In addition, user studies demonstrate strong consistency between CREval's automated metrics and human judgments. Therefore, CREval provides a reliable foundation for evaluating image editing models on complex and creative image manipulation tasks, and highlights key challenges and opportunities for future research.