Saved in:
Bibliographic Details
Main Authors: Tai, Ying, Du, Nikai, Xie, Rui, Chen, Zhennan, Wang, Qian, Jiang, Zhengkai, Zhang, Kai, Yang, Jian
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.23461
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908854935093248
author Tai, Ying
Du, Nikai
Xie, Rui
Chen, Zhennan
Wang, Qian
Jiang, Zhengkai
Zhang, Kai
Yang, Jian
author_facet Tai, Ying
Du, Nikai
Xie, Rui
Chen, Zhennan
Wang, Qian
Jiang, Zhengkai
Zhang, Kai
Yang, Jian
contents In this paper, we present TextCrafter, a Complex Visual Text Generation (CVTG) framework inspired by selective visual attention in cognitive science, and introduce the "Text Insulation-and-Attention" mechanisms. To implement the selective-attention principle that selection operates on discrete objects, we propose a novel Bottleneck-aware Constrained Reinforcement Learning for Multi-text Insulation, which substantially improves text-rendering performance on the strong Qwen-Image pretrained model without introducing additional parameters. To align with the selective concentration principle in human vision, we introduce a text-oriented attention module with a novel Quotation-guided Attention Gate that further improves generation quality for each text instance. Our Reinforcement Learning based text insulation approach attains state-of-the-art results, and incorporating text-oriented attention yields additional gains on top of an already strong baseline. More importantly, we introduce CVTG-2K, a benchmark comprising 2,000 complex visual-text prompts. These prompts vary in positions, quantities, lengths, and attributes, and span diverse real-world scenarios. Extensive evaluations on CVTG-2K, CVTG-Hard, LongText-Bench, and Geneval datasets confirm the effectiveness of TextCrafter. Despite using substantially fewer resources (i.e., 4 GPUs) than industrial-scale models (e.g., Qwen-Image, GPT Image, and Seedream), TextCrafter achieves superior performance in mitigating text misgeneration, omissions, and hallucinations.
format Preprint
id arxiv_https___arxiv_org_abs_2503_23461
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Investigating Text Insulation and Attention Mechanisms for Complex Visual Text Generation
Tai, Ying
Du, Nikai
Xie, Rui
Chen, Zhennan
Wang, Qian
Jiang, Zhengkai
Zhang, Kai
Yang, Jian
Computer Vision and Pattern Recognition
In this paper, we present TextCrafter, a Complex Visual Text Generation (CVTG) framework inspired by selective visual attention in cognitive science, and introduce the "Text Insulation-and-Attention" mechanisms. To implement the selective-attention principle that selection operates on discrete objects, we propose a novel Bottleneck-aware Constrained Reinforcement Learning for Multi-text Insulation, which substantially improves text-rendering performance on the strong Qwen-Image pretrained model without introducing additional parameters. To align with the selective concentration principle in human vision, we introduce a text-oriented attention module with a novel Quotation-guided Attention Gate that further improves generation quality for each text instance. Our Reinforcement Learning based text insulation approach attains state-of-the-art results, and incorporating text-oriented attention yields additional gains on top of an already strong baseline. More importantly, we introduce CVTG-2K, a benchmark comprising 2,000 complex visual-text prompts. These prompts vary in positions, quantities, lengths, and attributes, and span diverse real-world scenarios. Extensive evaluations on CVTG-2K, CVTG-Hard, LongText-Bench, and Geneval datasets confirm the effectiveness of TextCrafter. Despite using substantially fewer resources (i.e., 4 GPUs) than industrial-scale models (e.g., Qwen-Image, GPT Image, and Seedream), TextCrafter achieves superior performance in mitigating text misgeneration, omissions, and hallucinations.
title Investigating Text Insulation and Attention Mechanisms for Complex Visual Text Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.23461