Urban Mining Reloaded: Scanning Building Elements with Gaussian Splatting

Ferschin, Peter and Kovacs, Balint and Erb, Ingrid (2026) Urban Mining Reloaded: Scanning Building Elements with Gaussian Splatting. EVERYBODY PLANS ... SOMETIMES. Cherish Heritage, Plan Now, Create a Better Future! Proceedings of REAL CORP 2026, 31st International Conference on Urban Development, Regional Planning and Information Society. pp. 995-1002. ISSN 2521-3938

Text (Urban Mining Reloaded: Scanning Building Elements with Gaussian Splatting) CORP2026_108.pdf - Published Version Download (1MB)

Abstract

The construction sector is one of the largest consumers of raw materials. This consumption can be reduced through reuse and recycling of building materials and elements that have reached their end-of-life. However, there is a lack of information on the existing building stock as well as missing information on re-usable building elements suitable to be integrated into architectural design. This paper addresses both areas by presenting methods to estimate the availability of urban materials from city models and by exploring novel approaches for scanning 3d elements using Gaussian splatting. We discuss practical experiments on case studies of buildings in the city of Vienna, Austria. To estimate the material stock, 3d models (LOD 2.1 CityGML) were enriched with GIS data on building periods and typical basement heights from those periods – information which are not included in the surface geometry of the city models. From the evaluation of case studies that were surveyed with detailed representation of geometry and materials and subsequently converted into BIM models, material intensities were determined for specific building types. These data allow rough estimations of the material stock for similar buildings based on gross volume calculations. To provide resources for the re-use of building elements, Gaussian splatting as a digitalization technique was chosen to reduce the deficits of traditional methods like laser scanning and photogrammetry, which often fail when capturing objects consisting of transparent or reflective materials (e.g. glass). We present the integration of 3d objects represented as Gaussian splats into an application for early architectural design (MR.Sketch) and describe the process of converting Gaussian splats into point clouds for the use in BIM environments.

Item Type:	Article
Uncontrolled Keywords:	urban mining, gaussian splatting, building stock, planning, cities
Subjects:	T Technology > T Technology (General) T Technology > TH Building construction
Depositing User:	The CORP Team
Date Deposited:	09 Apr 2026 21:00
Last Modified:	09 Apr 2026 21:00
URI:	http://repository.corp.at/id/eprint/1399

Actions (login required)

View Item