Digital City Models · Geoinformatics · Automation

CityGML workflows for Blender, machine learning and 3D geospatial data

Selected development projects address the automated generation, modelling and processing of digital city models. The focus is on open standards, reproducible geospatial workflows and specialised tools for CityGML.

Selected Work

Projects

Six development areas combine CityGML, Blender, parametric modelling, remote-sensing data, machine learning and standardised Web Processing Services. Each project explicitly states its development status and the nature of its visual presentation.

Functional prototypeTitle graphic: schematic

CityGML Import and Export in Blender

The Blender plugin supports georeferenced import, semantic editing and export of CityGML 2.0 and 3.0. It includes file and 3DCityDB workflows, ModelTyper, the Openings Cutter, export and validation functions.

Blender CityGML Importer/Exporter on GitHub Repository, releases and source code

I developed this plugin substantially as part of my work at Virtual City Systems. I now maintain the current open-source continuation independently.

Read the Virtual City Systems project article
CityGMLBlenderPython
Functional prototypeTitle graphic: schematic

Blender Building Detailer – Parametric Building Details

The Blender plugin creates parametric building details on selected roof and façade surfaces, including dormers, chimneys, balconies, bay windows and roof parapets. It supports placement, snapping, geometry checks and semantic finalisation as merged geometry or a BuildingPart.

Blender Building Detailer on GitHub Repository, releases and source code
BlenderBuildingPartsCityGMLLAS
Experimental workflowSchematic · not a benchmark

Opening Detection and Generation Using Machine Learning

An experimental development workflow investigates the detection of windows and doors in building façades. Detections are geometrically refined and prepared as a basis for CityGML openings. Performance figures are reported only for clearly delimited test datasets and defined evaluation metrics.

Machine LearningOpeningsCityGML
Functional prototypeTitle graphic: schematic

Automated Texturing of CityGML City Models

The functional prototype projects georeferenced aerial imagery onto CityGML surfaces, evaluates suitable camera perspectives and generates texture assignments and atlases. The focus is on reproducible processing steps and traceable image assignment.

PhotogrammetryTexturingCityGML
Development prototypeTitle graphic: schematic

Automated Generation of CityGML Building Objects

The development prototype combines building footprints, digital terrain models and classified LAS point clouds. It derives georeferenced building geometry with height determination, roof reconstruction and semantic surface assignment.

LASDTM3D Reconstruction
Functional prototypeTitle graphic: schematic

Web Processing Services for 3D Geospatial Data in Blender

A Blender-based client integrates standardised OGC Web Processing Services into 3D GIS workflows. Geometry, geospatial data and parameters are submitted to PyWPS services in a controlled process; generated 3D geometry, point clouds, raster and vector data can be processed further in Blender while preserving georeferencing.

Blender WPS Client on GitHub Repository, releases and source code
WPSPyWPSBlender3D GIS

Request temporary demo access

From Source Data to City Model

A Shared Technical Focus

The projects share one objective: to automate complex 3D geospatial processing steps in a transparent and reproducible way. Georeferencing, semantic structure, exchange formats, error handling and operational constraints are considered together.