A Taxonomy of Object/Building Oriented Viewpoint Selection Methods in Architecture: A Systematic Review

This study systematically examines 3D viewpoint selection techniques, focusing on object- and building-oriented methods. The research aims to fill the research gap created by the fact that different disciplines in literature (computer graphics, architecture, and urban planning) have addressed viewpoint selection in isolation, and to integrate these methods under a holistic taxonomy. Conducted in accordance with PRISMA guidelines, the methodology involved a search of the Web of Science, Scopus, and Google Scholar databases covering a broad time span from 1988 to 2025. The search utilized the keywords “viewpoint research method”, “viewpoint selection algorithms/methods”, “classifications of viewpoint selection methods” and “viewpoint selection”; out of over 7,000 sources, 38 foundational studies with the highest methodological suitability were included in the detailed analysis. The methods were classified into six main categories: geometric, aesthetic, visual features, semantic, deep learning, and architecture/urban model based. The findings indicate that geometric and entropy-based methods dominate in object-oriented approaches, while GIS (Geographic Information Systems)-based spatial analyses and GPU-accelerated real-time visibility calculations take precedence in building-oriented approaches. The analytical value provided by this proposed classification lies in its clearly identifying the most appropriate computational strategy for each domain by differentiating methods based on the scale of application (from individual objects to urban fabrics). This study contributes to literature by providing a fundamental reference point for viewpoint determination systems in hybrid fields such as smart cities and interactive digital museology.