MajutsuCity：语言驱动的美学自适应城市生成，具备可控3D资产与布局
MajutsuCity: Language-driven Aesthetic-adaptive City Generation with Controllable 3D Assets and Layouts
将城市表示为可控布局、资产与材质的组合，并通过语言智能体支持对象级编辑，为3D场景生成提供了可复用的结构化表示与编辑范式；同时提出的结构一致性、材质保真度等评测指标可用于评估其他3D生成方法。
原文依据 MajutsuCity represents a city as a composition of controllable layouts, assets, and materials
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论生成逼真的3D城市对于世界模型、虚拟现实和游戏开发至关重要，理想的城市场景需同时满足风格多样性、细粒度和可控性。然而，现有方法难以在基于文本生成提供的创作灵活性与显式结构表示支持的对象级可编辑性之间取得平衡。我们提出MajutsuCity，一个自然语言驱动且美学自适应的框架，用于合成结构一致且风格多样的3D城市场景。MajutsuCity将城市表示为可控布局、资产和材质的组合，并通过四阶段流程运行。为将可控性扩展到初始生成之外，我们进一步集成MajutsuAgent，一个交互式语言引导编辑智能体，支持五种对象级操作。为支持照片级真实感与可定制场景合成，我们还构建了MajutsuDataset，一个高质量多模态数据集，包含2D语义布局与高度图、多样3D建筑资产以及精选PBR材质和天空盒，每项均配有详细标注。同时，我们开发了一套实用评估指标，涵盖结构一致性、场景复杂度、材质保真度和光照氛围等关键维度。大量实验表明，MajutsuCity相比CityDreamer将布局弗雷歇初始距离降低83.7%，相比CityCraft降低20.1%。我们的方法在所有AQS和RDR分数上排名第一，以明显优势超越现有方法。这些结果证实MajutsuCity在3D城市生成的几何保真度、风格自适应性和语义可控性方面达到了新的最优水平。我们期望该框架能为3D城市生成研究启发新途径。项目页面：https://longhz140516.github.io/MajutsuCity/
英文原文摘要 1882 字符
Generating realistic 3D cities is fundamental to world models, virtual reality, and game development, where an ideal urban scene must satisfy both stylistic diversity, fine-grained, and controllability. However, existing methods struggle to balance the creative flexibility offered by text-based generation with the object-level editability enabled by explicit structural representations. We introduce MajutsuCity, a natural language-driven and aesthetically adaptive framework for synthesizing structurally consistent and stylistically diverse 3D urban scenes. MajutsuCity represents a city as a composition of controllable layouts, assets, and materials, and operates through a four-stage pipeline. To extend controllability beyond initial generation, we further integrate MajutsuAgent, an interactive language-grounded editing agent that supports five object-level operations. To support photorealistic and customizable scene synthesis, we also construct MajutsuDataset, a high-quality multimodal dataset containing 2D semantic layouts and height maps, diverse 3D building assets, and curated PBR materials and skyboxes, each accompanied by detailed annotations. Meanwhile, we develop a practical set of evaluation metrics, covering key dimensions such as structural consistency, scene complexity, material fidelity, and lighting atmosphere. Extensive experiments demonstrate MajutsuCity reduces layout FID by 83.7% compared with CityDreamer and by 20.1% over CityCraft. Our method ranks first across all AQS and RDR scores, outperforming existing methods by a clear margin. These results confirm MajutsuCity as a new state-of-the-art in geometric fidelity, stylistic adaptability, and semantic controllability for 3D city generation. We expect our framework can inspire new avenues of research in 3D city generation. Our project page: https://longhz140516.github.io/MajutsuCity/