Enregistré dans:
Détails bibliographiques
Auteurs principaux: He, Jun, Ye, Junyan, Huang, Zilong, Jiang, Dongzhi, Zhang, Chenjue, Zhu, Leqi, Zhang, Renrui, Zhang, Xiang, Li, Weijia
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2602.01756
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908805035458560
author He, Jun
Ye, Junyan
Huang, Zilong
Jiang, Dongzhi
Zhang, Chenjue
Zhu, Leqi
Zhang, Renrui
Zhang, Xiang
Li, Weijia
author_facet He, Jun
Ye, Junyan
Huang, Zilong
Jiang, Dongzhi
Zhang, Chenjue
Zhu, Leqi
Zhang, Renrui
Zhang, Xiang
Li, Weijia
contents While text-to-image generation has achieved unprecedented fidelity, the vast majority of existing models function fundamentally as static text-to-pixel decoders. Consequently, they often fail to grasp implicit user intentions. Although emerging unified understanding-generation models have improved intent comprehension, they still struggle to accomplish tasks involving complex knowledge reasoning within a single model. Moreover, constrained by static internal priors, these models remain unable to adapt to the evolving dynamics of the real world. To bridge these gaps, we introduce Mind-Brush, a unified agentic framework that transforms generation into a dynamic, knowledge-driven workflow. Simulating a human-like 'think-research-create' paradigm, Mind-Brush actively retrieves multimodal evidence to ground out-of-distribution concepts and employs reasoning tools to resolve implicit visual constraints. To rigorously evaluate these capabilities, we propose Mind-Bench, a comprehensive benchmark comprising 500 distinct samples spanning real-time news, emerging concepts, and domains such as mathematical and Geo-Reasoning. Extensive experiments demonstrate that Mind-Brush significantly enhances the capabilities of unified models, realizing a zero-to-one capability leap for the Qwen-Image baseline on Mind-Bench, while achieving superior results on established benchmarks like WISE and RISE.
format Preprint
id arxiv_https___arxiv_org_abs_2602_01756
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Mind-Brush: Integrating Agentic Cognitive Search and Reasoning into Image Generation
He, Jun
Ye, Junyan
Huang, Zilong
Jiang, Dongzhi
Zhang, Chenjue
Zhu, Leqi
Zhang, Renrui
Zhang, Xiang
Li, Weijia
Computer Vision and Pattern Recognition
While text-to-image generation has achieved unprecedented fidelity, the vast majority of existing models function fundamentally as static text-to-pixel decoders. Consequently, they often fail to grasp implicit user intentions. Although emerging unified understanding-generation models have improved intent comprehension, they still struggle to accomplish tasks involving complex knowledge reasoning within a single model. Moreover, constrained by static internal priors, these models remain unable to adapt to the evolving dynamics of the real world. To bridge these gaps, we introduce Mind-Brush, a unified agentic framework that transforms generation into a dynamic, knowledge-driven workflow. Simulating a human-like 'think-research-create' paradigm, Mind-Brush actively retrieves multimodal evidence to ground out-of-distribution concepts and employs reasoning tools to resolve implicit visual constraints. To rigorously evaluate these capabilities, we propose Mind-Bench, a comprehensive benchmark comprising 500 distinct samples spanning real-time news, emerging concepts, and domains such as mathematical and Geo-Reasoning. Extensive experiments demonstrate that Mind-Brush significantly enhances the capabilities of unified models, realizing a zero-to-one capability leap for the Qwen-Image baseline on Mind-Bench, while achieving superior results on established benchmarks like WISE and RISE.
title Mind-Brush: Integrating Agentic Cognitive Search and Reasoning into Image Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.01756