Point Cloud to Revit Model Automation
When you need to capture as-built conditions there are multiple methodologies to consider. A few of the high-tech options growing in popularity are LiDAR scanning & Photogrammetry to assist in the collection of data. Both methods produce point clouds with vectorized data which means everything can be measured.
Scan Objects Matter
For some projects, a point cloud to verify measurements is all that is needed but many want to take that information and create a 3D model to serve as a starting point for a new design or construction model.
Unfortunately, there is no “easy button” to convert a point cloud to a complete and perfect 3D model. There is, however, a solution that makes it easier by automating the extraction of some shapes. EdgeWise from ClearEdge3D can extract surfaces and cylinders from point clouds automatically and convert them to 3D Revit Families, 3D AutoCAD, or Plant 3D blocks.
For the partial easy button “EdgeWise” to work well, the field must be scanned properly. One example is a recent Scan-to-BIM project of a mechanical tunnel with an abundance of pipes stretching 200 meters between the main hospital and central plant. This was the perfect project to leverage EdgeWise to reduce the manual conversion effort to Revit. We learned a best practice along the way to get better automatic extraction results.
Overcoming Field Challenges
The 200-meter tunnel had a rack for pipes on one side, two deep and four high, very close together. Our field technician scanned the tunnel, stitched the scans together, and then ran it through EdgeWise. The results were not as great as expected. The scan did not extract many of the pipes against the wall.
By looking at the items that were extracted accurately, we could clearly see that the scans captured all the front pipes. The back pipes were missing, or a partial image. Looking closer at the point cloud data we saw that the front pipes were blocking the view of back pipes. We did not achieve the 30% coverage that is ideal to extract the cylinders automatically. The first set of scans were all done at the same height along the tunnel path. The lesson learned is to vary the scanner height at each location zig-zagging up and down to get better angles on the pipes in the back.
To improve our extraction results we went back and scanned the tunnel again, this time using a zigzag, up and down pattern which gave us more than 30% coverage of the pipe surfaces hidden in the back. We then added the additional scans to the initial point cloud and ran it through Edge Wise.