WZMH Architects

The point cloud dataset for the 375/394 University Avenue project was 84GB and needed to be accessed within the design software of five team members at WZMH and all of the client’s facility management team. An additional complication was that the facility management team did not have software to view and manipulate point cloud data.

The Solution – Convert Point Cloud into Classified Mesh Models

WZMH first adopted PointFuse and its mesh modelling approach to managing all its laser scanned information at the end of 2019. For WZMH, the time spent duplicating and cropping point clouds so that they could conform to the import restrictions of the existing and varied range of design software used by the team had the potential to cancel the efficiencies delivered by investing in scanning.

“After extensive testing of a range of point cloud management solutions, it was clear that meshes were the route that we needed to go” concluded Dixon.

On this project converting and seamlessly storing meshes in Naviswork’s NWC format, which is then in turn referenced by Revit as a coordinated model made most sense. While automatic mesh conversion was left to run overnight, , and then around an hour was spent setting-up Revit templates for the project in the morning, this coordinated mesh was essentially available in Revit for the design team to get working on straight away. When asked how long the process of getting the reality capture data into Revit would have taken by importing the point cloud into Revit using conventional methods, Dennis said:

“I don’t know if I would like to think! Perhaps between 3 to 4 hours for each license of Revit being used on the project. Across a design team of three people, these means that using the mesh generated by PointFuse saved between 9-12 hours of project time before the designers could even get to work.”

In the case of the 375/393 University Avenue project, original point cloud consumed 84GB of disk space. Once converted into the mesh model, the size of the scanned data was reduced to just 1% of its original size. To further optimize file sizes and the as-built data views used by the design team, Dixon used PointFuse to classify the mesh prior to making it available to the project team. The classes included envelope elements (walls, windows, doors, floors, ceilings), fixed and moveable furniture, filtered noise (such as, scanner refractions from a mirrored columns and a custom Diamond-Waffle pattern on the ceiling)  and unfiltered noise (artefacts such as people and plants). These classification layers were then turned on and off in the design, and 3D coordination software used on the project.

WZMH’s design team needed to undertake visualization and feasibility work using the Unreal Engine platform, for which a lightweight textured mesh in OBJ format was also exported from PointFuse. Working with a mesh that had existing furniture removed meant that scenarios could be played out that included inserting new elements and furniture into the existing scene.

This output really came into its own when it came to sharing with the client’s facility management team who had no means to view the point cloud data, and especially not a point cloud dataset of this size. PointFuse does not subsample image textures to fit to the mesh, so it provides a high degree of realism for VR applications. Providing the data in OBJ and viewing in MeshLab, the client was able to walk through the site for themselves and be fully engaged with WZMH’s design team and visualize site conditions in a more complete manner than existing CAD drawings could offer.

Outcomes and Return on Investment (ROI)

“Keeping our data in the NWC format meant that we reduced the size of the laser scanned baseline data used on this project by a massive 99%. Our multi-person design team were able to simultaneously access all the scan data on a project at once and without causing network issues” explained Dixon. “On top of minimizing the volume of data that needed to be managed on this project there is interoperability with the range of software tools that we use, all without the potential overhead of third party plugins.”

Because you can simply ‘turn-off’ the noise in the scan data, PointFuse’s classified mesh data structure enabled the teams to only focus on the structural elements. Indeed, the design team were able to operate far more efficiently, eliminating the significant time that would otherwise be spent cropping and duplicating data, meaning they could concentrate on applying their professional expertise to the design of this project. Finally, reducing the volume of data being passed back and forth across WZMH’s network brought operational efficiencies. All these factors lead to competitively costed projects and the delivery of extra value.

For Dennis though, the outcomes were much simpler than that,“working with PointFuse’s classified meshes was essentially the different between being able to collaborate with our client, or not. We needed our reality capture data to be more portable and viewable against our future designs. There was otherwise no means at all for our client to view a point cloud data of this volume”. Indeed, using such as compressed mesh was also the only way the WZMH team could work with the data within their visualization platform”. Dennis continued, “At no extra cost to the project, we could provide a clean, uncluttered mesh enabling them to fully interrogate their space, especially with respect to the impact that the existing multilevel nature of the space has on egress through the property; and ensure that our design team could return a product that exceeded their requirements.”

Now, WZMH corporate policy is to manage all reality capture data as a mesh. With the conservation of project time being so important, immediately removing the requirement to continuously duplicate and crop data to fit with downstream requirements; not only brings tangible cost savings, but means that the AEC professional and their client can keep their focus on interpreting the data and not on needing to prepare the data. At WZMH, with their reality capture data now managed as a classified mesh, primarily alongside other site data in the Navisworks 3D Coordination environment, all of their reality capture data is not only now used in its entirety in the Revit design environment, but the same base dataset is also easily imported into the multitude of other visualization environments used by the team and their clients.