CoEditor++: Instruction-based Visual Editing via Cognitive Reasoning

TL;DR

CoEditor++ is a training-free, cognitively structured framework for instruction-based image editing that decomposes editing into two cognitive stages, achieving state-of-the-art results in visual consistency and interpretability without additional training.

Contribution

Proposes CoEditor++, a novel open-source, training-free framework that decomposes image editing into cognitive stages, enhancing robustness and interpretability in instruction-based visual editing.

Findings

Achieves state-of-the-art performance on SmartEdit and AltBear benchmarks.

Maintains high visual consistency compared to trained models.

Outperforms closed-source models like Nano Banana Pro and GPT-4o in visual quality.

Abstract

Recent advances in large multimodal models (LMMs) have enabled instruction-based image editing, allowing users to modify visual content via natural language descriptions. However, existing approaches often struggle with high-level semantic reasoning and visual consistency, particularly under ambiguous or complex instructions. To address these challenges, we propose CoEditor++, a cognitively structured, training-free framework that decomposes editing into "what to edit" and "how to edit" through two cognitive stages with a reflective self-selection mechanism, enabling robust, fine-grained, and interpretable editing. Built entirely from open-sourced components, CoEditor++ requires no additional training or fine-tuning, ensuring transparency and cross-domain applicability. We evaluate CoEditor++ on SmartEdit, a widely used benchmark for general editing, and AltBear, a privacy and compliance-oriented benchmark. Experimental results show that CoEditor++ achieves state-of-the-art performance in both general editing and responsible editing tasks compared with open-sourced models that require training on specialized editing datasets maintaining significantly higher visual consistency. When compared with closed-source models such as Nano Banana Pro or GPT-4o, CoEditor++ preserves comparable instruction following while still substantially outperforming them in visual consistency. Extensive ablation studies confirm that the effectiveness of CoEditor++ benefits from its structured cognitive design rather than any specific model component. Our findings suggest the potential toward cognitive-centric instruction-based image editing.