WorldEdit: Towards Open-World Image Editing with a Knowledge-Informed Benchmark

TL;DR

This paper introduces WorldEdit, a new dataset and evaluation framework for open-world image editing that emphasizes causal reasoning and real-world knowledge, addressing limitations of existing models in implicit instruction understanding.

Contribution

The paper presents WorldEdit, a novel dataset and benchmark designed to improve implicit, causal, and knowledge-driven image editing capabilities in models.

Findings

WorldEdit enables models to better handle implicit and causal instructions.

Fine-tuning with WorldEdit improves performance on causal editing scenarios.

Models trained with WorldEdit outperform existing systems in instruction following and knowledge plausibility.

Abstract

Recent advances in image editing models have demonstrated remarkable capabilities in executing explicit instructions, such as attribute manipulation, style transfer, and pose synthesis. However, these models often face challenges when dealing with implicit editing instructions, which describe the cause of a visual change without explicitly detailing the resulting outcome. These limitations arise because existing models rely on uniform editing strategies that are not equipped to handle the complex world knowledge and reasoning required for implicit instructions. To address this gap, we introduce \textbf{WorldEdit}, a dataset specifically designed to enable world-driven image editing. WorldEdit consists of high-quality editing samples, guided by paraphrased instructions that align with real-world causal logic. Furthermore, we provide \textbf{WorldEdit-Test} for evaluating the existing model's performance on causal editing scenarios. With WorldEdit, we use a two-stage training framework for fine-tuning models like Bagel, integrating with a causal verification reward. Our results show that the proposed dataset and methods significantly narrow the gap with GPT-4o and Nano-Banana, demonstrating competitive performance not only in instruction following but also in knowledge plausibility, where many open-source systems typically struggle.