ReasonEdit: Towards Reasoning-Enhanced Image Editing Models

Fukun Yin, Shiyu Liu, Yucheng Han, Zhibo Wang, Peng Xing, Rui Wang, Wei Cheng, Yingming Wang, Aojie Li, Zixin Yin, Pengtao Chen, Xianfang Zeng, Gang YU, Daxin Jiang

Keywords: Unified model; Understand; Image Edit

Abstract:

Recent advances in image editing models have shown remarkable progress. A common architectural design couples a multimodal large language model (MLLM) encoder with a diffusion decoder, as seen in systems such as Step1X-Edit and Qwen-Image-Edit, where the MLLM encodes both the reference image and the instruction but remains frozen during training. In this work, we demonstrate that unlocking the reasoning capabilities of MLLM can further push the boundaries of editing models. Specifically, we explore two reasoning mechanisms, thinking and reflection, which enhance instruction understanding and editing accuracy.
Based on that, our proposed framework enables image editing in a thinking–editing–reflection loop: the thinking mechanism leverages the world knowledge of MLLM to interpret abstract instructions, while the reflection reviews editing results and automatically corrects unintended manipulations. Extensive experiments demonstrate that our reasoning approach achieves significant performance gains over the base model, with improvements of ImgEdit (+0.7%), GEdit (+1.4%), and Kris (2.5%), and also outperforms previous open-source methods on both Kris and ImgEdit.

Primary Area: generative models

Submission Number: 1891