This paper explores the utilization of diffusion models and textual guidance for achieving localized editing of building facades, addressing the escalating demand for sophisticated editing methodologies in architectural design and urban planning. Leveraging the robust generative capabilities of diffusion models, this study presents a promising avenue for realistically synthesizing and modifying architectural facades. Through iterative diffusion and text descriptions, these models adeptly capture both the intricate global and local structures inherent in architectural facades, thus effectively navigating the complexity of such designs. Additionally, the paper examines the expansive potential of diffusion models in various facets, including the generation of novel facade designs, the enhancement of existing facades, and the realization of personalized customization. Despite their promise, diffusion models encounter obstacles such as computational resource constraints and data imbalances. To address these challenges, the study introduces the innovative Blended Latent Diffusion method for architectural facade editing, accompanied by a comprehensive visual analysis of its viability and efficacy. Through these endeavors, we aims to propel forward the field of architectural facade editing, contributing to its advancement and practical application.

翻译：本文探索了利用扩散模型与文本引导实现建筑立面局部编辑的方法，以应对建筑设计与城市规划领域对精细化编辑技术日益增长的需求。凭借扩散模型强大的生成能力，本研究为建筑立面的逼真合成与修改提供了可行路径。通过迭代扩散过程与文本描述，这些模型能够精准捕捉建筑立面固有的复杂全局与局部结构，从而有效应对此类设计的复杂性。此外，本文还考察了扩散模型在多种场景中的广阔应用潜力，包括新立面设计生成、现有立面增强以及个性化定制实现。尽管前景广阔，扩散模型仍面临计算资源限制与数据不均衡等挑战。为应对这些问题，本研究创新性地提出了用于建筑立面编辑的混合潜扩散方法，并通过全面的视觉分析验证了其可行性与有效性。通过这些工作，我们旨在推动建筑立面编辑领域的发展，助力其进步与实际应用。