The world could only watch in horror as Paris’s Notre Dame Cathedral burned in April 2019. The tower’s iconic spire was lost, but so too was much of the cathedral’s wooden framework, built from some 51 acres of forest, cut down almost a millennium ago. There aren’t adequate supplies of wood or trees tall enough to replace the framework faithfully, but it’s not all bad news: thanks to the efforts of a dedicated historian, Andrew Tallon, there is a point cloud—more than a billion data points that map the cathedral as it was in 2010, before the fire.
Motivated by a lifelong passion for Notre Dame, Tallon set out to capture it in all its beauty. Using laser scanners positioned on tripods, he captured every aspect of the building — sometimes putting concerns for his own safety aside to scale the building’s high ledges. He also took panoramic colour photos and used them to add RGB data to the point cloud.
The restoration is now fully underway, with a promise to combine contemporary and traditional architecture, and it would be wonderful to think that the final design will feature more than a nod to that original 1160 framework. If it does, it will be thanks in part to that point cloud and, more than a decade later, state-of-the-art point cloud mapping software to interpret its data in a way that renders it usable to architects, engineers, and restorers.
Benefits of point clouds
Mathematical objects can be described using formulae; architects’ drawings offer a sanitized view of how a building might look; photographs can capture the beauty of an old building. But none of these can capture all the detail and nuance of a building that has weathered and aged over time.
Point clouds capture the x, y, and z coordinates of every single point in an object or building. This doesn’t just mean that every brick is captured and recorded, but every ding, dent, and dimple gets a mention too. Contextual information, such as RGB data, helps build a repository of information that can be used to build a stunningly detailed 3D model.
It used to be the case that the model was built by hand, by specialist CAD technicians, but the process can now be automated using specialist software that constructs a realistic copy of the building virtually, using the point cloud and any contextual data captured during the scan.
This GIF of Beit Ghazeleh, an endangered heritage site in Aleppo, Syria, was built using specialist point cloud CAD software from a sample of a survey that captured 1.2 billion data points.
By loading the video, you agree to YouTube’s privacy policy.
Learn more
Barriers and challenges
So point clouds can store the minutest detail of a historic building, allowing it to be reproduced faithfully at some point in the future—it’s like capturing and preserving its DNA. But with that detail comes their biggest drawback: the files are huge, making it incredibly difficult to save, share, and use them.
It also takes a huge amount of skill to determine what the points actually represent and to connect the points—even then, human error can introduce inaccuracies. And you’re going to want people from different disciplines to work on the same data, so it’s vital that it can be used seamlessly across teams and software packages. Cloud solutions allow users to access a single copy of the file and to be confident that they are working on the latest version. Access protocols ensure that only one user can save changes at any one time.
These issues are common to any large-scale building project, but restoration brings an extra challenge: how to capture the ‘as-was’ condition of the building, before disaster struck, not just how it is now. Of course, we can’t travel back in time to get that ‘as-was’ picture, but what we can do is make the archivist’s job easier, with software that’s now far simpler to use, so that capturing a building’s data is something that can be done on a semi-regular basis, rather than being a major project.
The mesh: every point is related
Every point in the universe has a relative position to every other point, and we can describe that relationship in terms of vectors. By converting those point clouds into a ‘mesh’ of interconnected points, we can strip out redundant data and start to reduce file sizes—the bigger the triangle between three points, the smaller the file. The key here is that we’re only losing duplicated data — and what we sometimes refer to as noise (people and elements that are not required in the final model) — but none of the detail that makes the final 3D model so powerful.
Of course, if you built the mesh manually, it would still be prone to human error but there are now powerful point cloud software applications that can do the job quickly and accurately, doing away with the need for an experienced CAD technician and reducing file sizes from gigabytes to megabytes.
A three-step process for building restorations
The story doesn’t end with the mesh model. The mesh itself is a means to an end—a way of mapping the point cloud into a much smaller yet equally detailed file that can be used by multiple people across different as-built BIM systems.
An everyday building transformation would start with a point cloud to capture the ‘as-built BIM’ or ‘as-is’ position, representing a current snapshot of what’s real, from which architects would overlay the final design.
When it comes to preserving and restoring our architectural heritage however, there’s a vital extra pre-emptive step, and that’s to capture the ‘as-was’ by taking regular point cloud ‘backups’ and storing them safely. LiDAR sensors, or other handheld or body-worn scanners enable users to take scans, but larger and more complex scans have usually been handled by third-party suppliers so far, who might even deploy drone technology to get detailed scans of tall buildings.
Mesh software not only acts as a bridge between the data-heavy point clouds and the final design software, but can also provide a useful archiving tool, should the worst happen.
Author Bio
Mark Senior is a business director of PointFuse, a leading AEC software. He has been involved with PointFuse since its conception, shaping its development from bleeding-edge technology to the successful commercial solution it is today.
