I was going to start this piece by talking about “the humble point cloud,” but “humble” is something point clouds most definitely are not! They’re massive files, proudly offering a wealth of data for any AEC software specialist able to interpret the data successfully. But what’s the future of point clouds – and the 3D models that are so often the results of working with point clouds?
A Google search of “the history of point clouds” returns surprisingly few results, so let’s start there.
The history of point clouds
Point clouds have been around as long as 3D LiDAR scanners – since the 1960s. Over time, point clouds have grown in size and in detail, as the scanning technology that creates them has grown more sophisticated.
Initially mainly used by the military and space agencies for scanning terrain, by the 1990s point clouds were being used for a variety of industrial applications, including architecture, engineering, and archaeology. Just about anything can be scanned and turned into a point cloud, from landscapes, to buildings, to tiny archaeological artifacts.
What can be done with point clouds?
The applications of point clouds are, in 2021, many and varied:
- AEC software applications can use point cloud data to help architects and construction professionals incorporate accurate real-world data into as-built BIM
- Archaeologists can analyse point clouds of terrain to uncover hidden signs of human settlements, identifying new dig sites. They can also capture point clouds of artifacts, which can then be used to create digital replicas for archiving or for further research.
- Medical sciences professionals can use point clouds of human beings to design bespoke treatments or prosthetics for patients.
- Entertainment companies can use point clouds of real-world spaces to produce ever more accurate and realistic digital effects for use in games, movies and TV.
So that’s what’s happened with point clouds up until now. But what does the future hold?
The future of point clouds: overcoming limitations
The main advancements in point clouds since their inception in the 60s, is that their size has grown. And though they have a wide variety of uses, there are some limits with point clouds that are holding users back:
- Point clouds are so large that they require cutting-edge computer technology to successfully register and display them, which is expensive.
- The process of turning point clouds into usable 3D objects for AEC software and other applications is convoluted and highly manual.
There are already some exciting innovations that are helping to address these problems (including our own efforts here at PointFuse):
Cloud storage is helping overcome file sizes
The challenge of space has always been there with point clouds – as computers have grown in storage capacity and processing power, point clouds have grown in size to keep pace. So now, companies are looking at leveraging the cloud to make it easy to store and share point cloud data with their project teams.
For instance, construction firms often have multiple project teams who require access to the same as-built BIM model. Instead of everyone having their own copy of the point cloud, the point cloud is stored on a remote server, and the teams all access the same data via the cloud.
There are still some issues around latency to overcome, but as network infrastructure continues to improve – and in particular as 5G becomes ubiquitous – these limitations should disappear.
Adding intelligence to point cloud data
This is the thornier challenge to solve. Point cloud data is dense – a scan of a building will contain walls, doors, lights, and furniture – but also people, random objects on people’s desks, and other irrelevant data. Currently the process of deciding what data is important – and classifying it – is a manual one.
Some researchers are looking at ways to add more data into the points of the point cloud itself – an interesting solution, though one that risks increasing the data challenge I mentioned above. Another solution is to look at creating meshes from point cloud data. This is already done in various AEC software applications, but usually with a view to creating a specific visualisation. At PointFuse, though, we’ve experimented with using the meshing process to add intelligence to the data – classifying the point cloud data automatically for the user.
A world of innovation awaits
I’m confident that the challenges around point cloud data are temporary. The pace of human innovation – particular in computer sciences – is accelerating daily, so it’s only a matter of time before point clouds can be leveraged to their fullest extent, by as many people as possible. It’s an effort I’m proud to be a part of at PointFuse – and I’m really excited to see how our industry continues to evolve.
About the author:
Mark Senior is a business director of PointFuse which specializes in converting point cloud to mesh using point cloud 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.