StoryState: 基于代理的状态控制用于一致且可编辑的故事书
StoryState: Agent-Based State Control for Consistent and Editable Storybooks
AI总结:
StoryState通过基于代理的状态控制实现多页故事书的一致性与编辑优化
摘要:
大型多模态模型已使一键式故事书生成成为可能,用户只需提供简短描述即可获得多页插图故事。然而,底层故事状态,如角色、世界设定和页面级对象,仍保持隐含状态,导致编辑粗粒度且往往破坏视觉一致性。我们提出了StoryState,一种基于代理的协调层,它在无需训练的文本到图像生成之上引入了显式且可编辑的故事状态。StoryState将每个故事表示为由角色表、全局设置和每页场景约束组成的结构化对象,并采用少量LLM代理来维护此状态并推导出1Prompt1Story风格的提示用于生成和编辑。完全通过提示操作,StoryState是模型无关的,兼容多种生成后端。系统层面的多页编辑任务实验表明,StoryState能够实现局部页面编辑,提高跨页一致性,减少意外更改、交互轮次和编辑时间,相比1Prompt1Story,同时接近Gemini Storybook的一键一致性。代码可在https://github.com/YuZhenyuLindy/StoryState获取。
英文摘要:
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
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。