The renovation of Ljubljana Castle is an ongoing project since 1963, financed by Ljubljana Castle Public Institution and carried out by the architectural studio Ambient. The latest renovation occurred at the castle’s wine cellar and its outer patio. A 3D model of the patio was made and further used for the visualisation of patio.
The Department of Geodetic Engineering, Faculty of Civil and Geodetic Engineering at the University of Ljubljana, measured the castle’s outer wall and terrain with a terrestrial laser scanner (TLS). To compare results, a point cloud using 3Dsurvey was computed.
As this part of the castle is not suitable for aerial images, ground photos were taken with a Google smartphone, at different vertical angles, with at least a 70% horizontal and vertical overlap. To reach a point cloud completeness, images had to be taken from different angles as the wall is in the shape of the letter L.
Project tools and specs:
Camera – Google smartphone, focal length 5 mm
Used images – 205 with at least 70% overlap
Number of gcps – 7 points snapped from TLS point cloud
Number of point cloud points – 11.887.764
All 205 images were taken with a focal length of 5mm. For comparison, laser scanned point cloud was used for the extraction of gcps. Suitable wall details as well as the porthole were used for sparse point cloud orientation. We achieved spatial accuracy better than 12mm.
3Dsurvey CAD tools were used to outline the cellar’s doors and doorstep. A 3D model was georeferenced to point the cloud’s location using exported CAD lines. The main goal of this task was to provide a background for the patio visualisation and for this reason, the 3D model was converted to a point cloud.
The castle’s point cloud was displayed together with the one made for the patio. As point clouds were of high density, they could be used as a realistic background for visualisation purposes.
To visualise the project better, both the 3Dsurvey point cloud and the one made with TLS were combined. A further detailed view shows slight differences of 2cm in the location of point clouds. Considering that the 3Dsurvey point cloud was computed from images, taken with smartphone camera, results show great potential for visualisation purposes.