Editing and Visualization of Large Point Clouds for Urban Planning Applications

Gorazd Gorup

Supervisor(s): Ciril Bohak, Žiga Lesar

University of Ljubljana

Abstract: Urban planning has become increasingly complex in recent years, necessitating more detailed and accurate representations of real-world environments to meet design objectives. While mesh and parametric geometry are dominant in current workflows, they often fail to capture essential contextual information. In contrast, point clouds have untapped potential for improving urban design processes. This paper addresses the challenge of interactively visualizing and editing large point cloud datasets from LiDAR scans, proposing methods that we implemented as a set of tools and utilities to enhance urban planning workflows. We extended the open-source 3D modelling software Blender with features for loading and visualizing large point clouds, and implementing levels-of-detail and classification mechanisms. We do so by adapting Blender's internal representations and functions to accommodate point geometry. We created tools for targeted editing operations and leverage Blender's Geometry Nodes to facilitate non-destructive, node-based editing for operations such as point copying, painting, and flattening -- basic tasks in urban design. We tested the rendering functionality of Blender's rendering engines to visualize different point attributes, such as colour, elevation, and classification. Finally, we evaluated our solution against Rhinoceros 8 and its Grasshopper extension, commonly used in urban planning, and identified areas where Blender's point cloud manipulation and visualization capabilities could be improved.
Keywords: Design, Geometry Processing, Human-Computer Interaction, Point-based Graphics, Rendering
Full text:
Year: 2025