Decoding Urban Landscapes Through Point Cloud Modelling: A New Teaching Approach

Back to overview

Background

My name is Carlos Bartesaghi Koc and I am lecturer in landscape architecture at The University of Adelaide, School of Architecture and Built Environment (SABE), Faculty of Engineering, Computing and Mathematical Sciences (ECMS). My research and teaching activities are focused on the spatial and temporal dynamics and interactions that occur in built and natural environments with a particular attention to urban climatology, remote sensing and urban forestry.

I transitioned into academia in 2015 when I commenced my PhD, and I first began using 3D point cloud models retrieved by aerial laser scanners or LiDAR (Light Detection and Ranging) technologies to inform my research on the thermal performance of green infrastructure on urban microclimate. However, it was not until 2020, when COVID-19 hit Australia and the rest of the world so hard, that I started using 3D point cloud models for teaching purposes. At that moment, there was an urgent need for developing innovative digital resources that could support in-class learning activities for remote (online) students.

The use of 3D point cloud models in combination with online-based platforms such as Sketchfab gave me the opportunity to develop creative ways to represent, analyse, interact and share models in a completely virtual environment. We commonly teach design studios and representation courses using analogue techniques (i.e., physical models, sketches, drawings) or digital 3D- and BIM-based models (i.e., in Revit, ArchiCAD, Rhinoceros, Grasshopper, Sketchup), so these new technologies supplemented our current learning activities with a range of new skills and techniques that students were not able to experience before. For instance, students in Australia and abroad were able to explore, manipulate, and examine study sites of varied scales, sizes, and complexity that were practically inaccessible at that time.

My interest in 3D laser scanning sparked further after visiting the Sketchfab site of the Chair of Landscape Architecture Christophe Girot (ETH Zürich) and getting inspired by the novel models developed by 3D guru Dr Philipp Urech (ETH Zürich). Communications and interactions between the three of us were possible thanks to Sketchfab, and this led to fruitful and ongoing scholarly collaborations.

Carlos Bartesaghi Koc lecturer

A new pedagogical approach to study the dynamic world

In a world of rapidly changing conditions, it is urgently required the application of novel and progressive pedagogical and scholarly approaches under new paradigms and lenses. Under this ‘systems-thinking’ approach, buildings and landscapes are the result of complex, dynamic and evolving phenomena that vary in space and time (or 4D). Hence, built and natural environments should be represented, examined and visualised as living organisms in constant process of self-organisation, adaptation and mutation.

Despite this, standard practices in architecture and landscape architecture professions heavily rely on inventories, abstract diagrams, representations and models that are inherently static and somehow rigid. However, 3D point cloud models can support the disaggregation (segmentation or de-construction) of buildings and landscapes into multiple parts through the implementation of innovative workflows. This process can facilitate the analysis and visualisation of complex aspects such as urban form modifications, synergistic relationships, environmental performance, succession and evolution, and environmental changes.

Flooding model of Brisbane developed by student Renfei Xu

Teaching & learning: an ongoing journey

With the knowledge in remote sensing and photogrammetry that I acquired as part of my research activities, I decided to re-envision various design studio and representation courses offered by our school in a way that integrates holistic perspectives and digitally-based techniques to represent, visualise, and evaluate the dynamically complex processes occurring in the real world.

The key aim of these new courses is to teach students how to construct digital (online-based) 3D or 4D models capable of communicating the spatial and temporal dynamics (properties) of sites, buildings or spaces using digital datasets, advanced visualisation (and sound) tools and remote sensing techniques, particularly laser-scanning technology (or LiDAR).

Before implementing 3D point cloud modelling and analysis into our lessons, the process began with a self-guided learning journey full of successes and failures. After endless hours and multiple attempts, several point cloud models were produced and slowly uploaded to my Sketchfab page. This collection has slowly grown over time and is expanding as I continue developing new lessons and learning from my own students.

Developing a workflow

Students were required to prepare a range of 3D point cloud models to represent the ever-changing and evolutionary aspects of existing local contexts (i.e., real sites, buildings, or outdoor/indoor spaces) or unbuilt design proposals of their own (i.e., building or landscape design projects from their design studios). To achieve this, it was essential to define and implement a clear research workflow to retrieve, map, visualise, and upload models to Sketchfab.

3D Workflow landscape architecture

3D workflow designed by student Thomas Doan

A common protocol was adopted which started with the retrieval of terrestrial or aerial laser scanner point clouds from publicly available databases such as ELVIS Geoscience portal or OpenTopography.org (among many others) or by using our recently acquired terrestrial laser scanner (Leica BLK360). Several techniques were taught and applied by students to deconstruct, segment, re-/colour, analyse and reassemble point clouds using open-source software such as CloudCompare and QGIS. In collaboration with a team of computational experts (Juliana Croffi and Victor Calixto), we developed a series of video tutorials to rapidly train students on these new methods. All these digital lessons are publicly available through my YouTube Channel.

For instance, point segmentation (based on scalar fields such as classifications, number of returns, or point intensity) allowed us to distinguish between different types of built and natural features, identify different plant species, or provide relief visualisations of the bare earth for further topographic analysis and simulations.

Urban tree species 3D

Different tree species and their origins identified for Adelaide CBD, project developed by student Matthew Foreman.

In order to colour point clouds, various techniques were explored and these included transferring RGB values from high-resolution aerial imagery (from Nearmap portal or Google Earth Pro) or custom imagery created with Adobe Photoshop using LASTools in QGIS. Other colouring methods such as scalar field interpolation from other entities (i.e., ESRIgrids, GeoTiffs or meshes) and ambient occlusion (using the plugin PCV/ShadeViz “Portion of Visible Sky“) were also applied in CloudCompare to transfer RGB values from simulations back into the point clouds.

Several analytical methods such as change detection were employed to examine the dramatic urban development, topographic alterations and land cover changes (i.e. deforestation) observed for a particular area over a period of time. The methods implemented to achieve these estimations include the cloud-to-cloud (C2C) distances or the M3C2 plugin available in CloudCompare. More sophisticated methods used in class have included (1) the simulation of surface runoff and water flows in Rhinoceros and Grasshopper (with Bison, Docofossor plugins) and the subsequent conversion of results into point clouds using Volvox plugin; (2) the simulation of outdoor thermal comfort using the SOLWEIG model available in the UMEP (Urban Multi-scale Environmental Predictor) plugin for QGIS and (3) the estimation of various inundation levels with a tool available in Plas.io.

Communicating in the digital era

The ease of embedding and displaying Sketchfab models on other online platforms such as MIRO, Milanote or Zoom has improved in-class discussion, participation, and engagement among students, tutors, and jurors irrespective of their geographic location. Some examples of students’ models and designs are available via the above hyperlinks. Another advantage of using Sketchfab is that it allowed us to embrace new digital technologies like Virtual Reality (VR) and Augmented Reality (AR), and provided our staff and students with fully interactive and immersive tools to communicate and disseminate our work not only within the school, but also within the broader community.

Milanote Sketchfab embed

Sketchfab models developed by students from DESST3515 – Representation III course embedded into the Milanote platform

The models developed in class have also allowed us to engage more effectively with prospective students through various outreach activities. For instance, in a recent Young Women in STEM (science, technology, engineering and mathematics) activity, several local high school students had the opportunity to visualise and interact with Sketchfab models to better understand the potential impacts of flooding, sea level rise and bushfires in the context of climate change and global warming.

In future, with the advent of new 3D remote sensing technologies, we will be able to conduct multifaceted performance-based analyses in a more rapid and precise way. With a better understanding and visualisation of these complex processes, researchers and educators will be able to communicate the scientific evidence in a more effective way so we can make more informed decisions to combat climate change and achieve a more sustainable future for all.

Researcher ProfileYouTube channelAcademic websiteResearchGateLinkedIn

About the author

Dr Carlos Bartesaghi Koc

Carlos is a Lecturer in Landscape Architecture at The University of Adelaide (Australia) where he uses remote sensing and computational modelling to investigate the environmental impacts of urban greenery and built form on climate, air quality, energy consumption and human health.



No Comments

    Leave a Reply

    Your email address will not be published. Required fields are marked *

    Related articles