twitter youtube facebook linkedin email
CONNECT WITH US:

Having issues dealing with very large point clouds in Revit or Navisworks or your other design and construction tools? In one webinar, SSOE Group demonstrates how they use ReCap to break up scans and use those simplified point clouds in Revit and Navisworks to represent existing conditions.

SSOE is an engineering, procurement, and construction management firm, serving the semiconductor, automotive, food, chemical, glass, manufacturing, healthcare, power, and general building industries. Founded in 1948 in Ohio, SSOE now has almost 30 offices around the world and has completed projects in 40 countries.

 

SSOE’s history of using reality capture

Prior to May 2014, SSOE only scanned a handful of projects per year. But in the next 12 months, they spent over half a million dollars scanning over 50 projects that represented a construction value of more than $750M.

What changed?

“One software package has changed the game for us … ReCap” says Mark LaBell, a BIM/CAD technical leader at SSOE. LaBell explains that prior to ReCap, Autodesk products did have a point cloud engine in many of its design applications, but the scanned data didn’t work very well in those applications because the sheer size of point clouds made them difficult to move or use across networks and seriously slowed down design applications.

But in 2014 they began using ReCap to break up and simplify point clouds. Since then, SSOE has adopted laser scanning as a standard of practice throughout its organization and as a standard of care for its clients.

You can click this link to replay the webinar on YouTube, but here are some of the highlights.

 

The importance of IT infrastructure

LaBell started that webinar with a brief explanation and tips for moving simplified point clouds between offices. After SSOE has broken up the scanned data into more digestible chunks, they share it with other SSOE offices or project partners or clients using either the ReCap web services or using their internal IT infrastructure comprised of especially good servers and Riverbed technology. LaBell gave an example of the importance of using high-end servers and network infrastructure, relating how recently they transferred a large point cloud (~100 gigabyte file) between offices across the country in less than an hour. But they also had to transfer it to a much closer office that didn’t have comparable high-end network infrastructure and that transfer took 18 hours.

LaBell also described the workstation hardware he suggests you use when viewing and manipulating the edited point clouds. For large projects—which he defined as projects with point clouds over 50 gigabytes—SSOE typically uses second, solid-state hard drives along with robocopy routines to mirror files across the network. For smaller projects—with points clouds less than 50 gigabytes—users generally work directly from the network.

 

Editing tips & tricks

LaBell began his discussion of how to edit scanned data by sharing a few tips/tricks they’ve learned:

  • To optimize performance, always edit scanned data on a SSD, local hard drive—never across a network.
  • To ensure that you’re selecting and clipping points correctly, use orthographic views and the Ortho Mode in ReCap.
  • For performance, always export a unified data set (ie combine the RCS files into a single file and then export). LaBell gave an example of a project where they didn’t do this, where they edited the scanned data, then chopped it up into individual RCS files. Exporting a single floor/RCS took 48 hours. LaBell then reunified all the individual RCS files and exported them, which only took 35 minutes.

LaBell explained that at the beginning of each project, they talk with the design team to identify project scope of work and deliverables, which forms the basis of how the point clouds will need to be used and manipulated. This will also help you identify what parts of the original scanned data can be edited out of and deleted from the point cloud to improve performance in other Autodesk applications. LaBell also recommends defining Worksets in your Revit models for each point cloud referenced, so users can quickly turn off and on specific point clouds. Similarly, define a master NWF in Navisworks for point clouds.

Simplifying point clouds

 

How to make a large project small

The rest of the webinar featured a live ReCap demonstration, showing how SSOE edited and broke apart a large point cloud of an industrial warehouse/manufacturing facility that SSOE was renovating. LaBell’s group edited down the scanned data set from the vendor to create 2D demolition plans and generate point clouds representing pre and post-demolition existing conditions that were used for reference and clash detection during the renovation design process.

The original scanned data of the entire building (~300,000 square feet) was about 250 gigabytes. The webinar featured one section of the building. Before editing, the point cloud of that section was 6.5 gigabytes. SSOE edited the original scanned data/point cloud by removing the floor, the roof deck, and the equipment, leaving just the utilities and the structure. The resulting point cloud used in the design process was only 1.2 gigabytes, representing close to a 75 percent reduction in file size.

LaBell begins the ReCap demo by opening up the point cloud of the building section in ReCap and creates scan regions for the roof, floor, and equipment. Then he starts identifying those regions, beginning with the floor. ReCap has a several methods to select and edit groups of points, including identifying a window, fence, or plane. LaBell uses the Plane selection tool to isolate the floor and roof of the building. He selects several points on the floor, and hits Enter to see how much of the floor was actually selected. He repeats this several times to completely select all the planes of the floor, and then assigns that selection to the Floor region and turns off the display of that region.

Simplifying point clouds with ReCap

He repeats this process for the roof, but because the roof has varying slopes between the main structural columns, he uses a side/orthogonal view to establish a clip region between two structural columns, turns off the display of everything outside of that clip region, and then goes back to a perspective view to select points on the roof between those main columns. He repeats this process for the other portions of the roof and then assigns those roof planes to the Roof region and turns off the roof.

Next he moves on to the equipment. He views the point cloud from side views and uses fences to select equipment between the major columns, sometimes temporarily selecting and blanking walls to ‘see through’ the point cloud. After selecting all the equipment, he assigns them to the Equipment region and turns off the equipment.

He’s left with just the structural and building systems. These are the elements of the existing building that will be incorporated within the renovation and need to be used as an existing conditions reference for design and project coordination. He mentions, but doesn’t demonstrate due to time considerations, that they use similar processes to select and isolate individual building systems such as lighting, electrical, sprinkler system, piping, ducting, and so on.

He unifies the scan data to create a single, unified RCS (see note about performance above) and then exports the individual ReCap regions. He imports the resulting RCS files into Navisworks, where the structural and building systems point clouds (the parts of the building that will remain) are combined with the Revit models of the design renovation and used for project coordination and clash detection.

Optimizing point cloud models with ReCap

 

Results

LaBell reports that by using these techniques to simply point clouds, the total size of the point clouds they referenced in Navisworks was just 32 gigabytes. This compares to the original point cloud of the project, which was 250 gigabytes … an 87 percent reduction!

Moreover, they did a clash detection test using the original point cloud, resulting in ~1,800 clashes between the design model and that unedited point cloud. The bulk of those clashes were between renovated building systems and parts of the building that were planned for demolition. By segmenting and simplifying the point clouds, LaBell’s team saved the designers the time associated with sorting through these 1,800 clashes to identify real clashes versus clashes with to-be demolished elements. This represents a substantial time savings for SSOE and a significant value-added benefit for its clients.

Optimizing ReCap

For more similar news around Reality Capture technologies, make sure to subscribe to our blog.

SIGN UP FOR BLOG UPDATES

TRY RECAP NOW

MEET BLK360 & RECAP 360 PRO

JOIN THE RECAP COMMUNITY

Twitter   Facebook   LinkedIn   Google+   YouTube   RSS

Featured Links

Reality Capture Guru

I'm the expert when it comes to Reality Capture! Want to learn more about it, subscribe to this blog and I'll share all my knowledge with you!

0 Comments;