Re-Align: Structured Reasoning-guided Alignment for In-Context Image Generation and Editing

TL;DR

Re-Align is a unified framework that enhances in-context image generation and editing by integrating structured reasoning-guided alignment, significantly improving understanding and faithful execution of user prompts.

Contribution

It introduces the In-Context Chain-of-Thought paradigm and an RL training scheme to better align reasoning with image generation in multimodal models.

Findings

Re-Align outperforms existing methods on image generation tasks.

The structured reasoning approach reduces confusion among reference images.

The RL scheme improves alignment accuracy between reasoning and generated images.

Abstract

In-context image generation and editing (ICGE) enables users to specify visual concepts through interleaved image-text prompts, demanding precise understanding and faithful execution of user intent. Although recent unified multimodal models exhibit promising understanding capabilities, these strengths often fail to transfer effectively to image generation. We introduce Re-Align, a unified framework that bridges the gap between understanding and generation through structured reasoning-guided alignment. At its core lies the In-Context Chain-of-Thought (IC-CoT), a structured reasoning paradigm that decouples semantic guidance and reference association, providing clear textual target and mitigating confusion among reference images. Furthermore, Re-Align introduces an effective RL training scheme that leverages a surrogate reward to measure the alignment between structured reasoning text and the generated image, thereby improving the model's overall performance on ICGE tasks. Extensive experiments verify that Re-Align outperforms competitive methods of comparable model scale and resources on both in-context image generation and editing tasks.