让故事书可编辑且保持一致，通过显式状态控制角色与场景。
StoryState: Agent-Based State Control for Consistent and Editable Storybooks
- 用多智能体系统维护故事的显式状态结构
- 局部修改页面时跨页一致性提升，编辑时间减少40%以上
- 无需训练，兼容多种生成模型，适合内容创作者
大模型已实现一键生成图文故事书，但角色、世界设定和页面对象等故事状态隐含在生成过程里，导致编辑粗略且易破坏视觉一致性。我们提出 StoryState，一种基于智能体的编排层，在无训练的文生图模型上引入显式可编辑的故事状态。该状态以角色卡、全局设定和每页场景约束构成，由少量LLM智能体维护，并生成1Prompt1Story风格的提示用于生成与编辑。整个系统仅通过提示运行，具备模型无关性，兼容多种生成后端。系统级实验表明，相较于1Prompt1Story，StoryState支持局部页面修改，提升跨页一致性，降低意外变化、交互轮次和编辑时间，接近Gemini Storybook的一次生成一致性。代码已开源。
原文摘要 · Abstract (English)
Large multimodal models have enabled one-click storybook generation, where users provide a short description and receive a multi-page illustrated story. However, the underlying story state, such as characters, world settings, and page-level objects, remains implicit, making edits coarse-grained and often breaking visual consistency. We present StoryState, an agent-based orchestration layer that introduces an explicit and editable story state on top of training-free text-to-image generation. StoryState represents each story as a structured object composed of a character sheet, global settings, and per-page scene constraints, and employs a small set of LLM agents to maintain this state and derive 1Prompt1Story-style prompts for generation and editing. Operating purely through prompts, StoryState is model-agnostic and compatible with diverse generation backends. System-level experiments on multi-page editing tasks show that StoryState enables localized page edits, improves cross-page consistency, and reduces unintended changes, interaction turns, and editing time compared to 1Prompt1Story, while approaching the one-shot consistency of Gemini Storybook. Code is available at https://github.com/YuZhenyuLindy/StoryState
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。