3-D scanning and the holographic landscape.
By Brian Barth
It’s been more than a decade since Google Earth put 3-D mapping in the hands of anyone with an Internet connection. Now, armchair map geeks can fly through the skyline of virtually any city in the world to check out, say, the architecture of the Louvre or take a virtual stroll through the Jardin des Tuileries using Google Street View. The ability to cost-effectively produce such imagery on a global scale stems in part from advances in 3-D scanning, a term of art that encompasses LiDAR (light detection and ranging), drone-based photography, ground-penetrating radar, and other advanced imaging technologies.
Three-dimensional scanning has become so inexpensive and user-friendly that design firms are starting to experiment with it. Architects and engineers use it to help create as-built drawings of bridges and buildings and for “clash detection” when designing additions or renovations of historic structures. Urban planners use it as a visualization tool when modeling different development scenarios. Anything that can be 3-D scanned can be 3-D printed, and numerous municipalities now have scale models of their cities in physical form, thanks to the intersection of the two technologies.
For landscape architects, it’s a way to generate a perfectly scaled 3-D model of any site, and thus a foundation on which to base design work—in theory, anyway. In practice, there are hurdles to integrating 3-D scanning technology into daily work flows, mainly owing to software constraints.
Surveying is one profession that has jumped fully into 3-D scanning, which has implications for landscape architects: The technology is a means to create the ultimate base map. Tate Jones, an Atlanta-based surveyor who specializes in 3-D scanning, says the technology is quickly replacing two-dimensional surveying methods because it is faster, more precise, and able to capture almost limitless detail about a site. The resulting “point cloud” of data, he says, is best imagined as a 3-D mesh consisting of millions of geo-referenced points—if not billions or trillions, depending on the scope of the project—over which a 360-degree photo of every inch of the site can be draped.
“It’s like having a 3-D hologram of the environment on your computer,” says Jones, who has completed scans for hundreds of sites over the past decade, including well-known Atlanta landscapes such as Centennial Olympic Park, Woodruff Park, and the Atlanta Athletic Club. Rather than make repeated site visits to consider a particular view or to plot out the precise location of exposed boulders or mature trees using a GPS device, designers can simply toggle through what amounts to their own personal Google Street View of the site to collect the information they need. Except 3-D scans are much more accurate than fuzzy Google Earth images when you zoom in—precision down to plus or minus a millimeter is possible.
Jones says: “If I’m in Atlanta, but my project happens to be in Jackson, Mississippi, and somebody says, ‘Okay, how many windows and doors are on the west side of that building?’ Well, you can just click there and see. It saves gasoline and plane tickets.”
No single type of 3-D scanning device can form a complete picture of the environment on its own. Different technologies are often used in conjunction, with the point clouds stitched together and viewable in a variety of software, including Autodesk’s Recap 360 and 3ds Max, and Agisoft PhotoScan. In fact, many CAD software packages can now view 3-D scanned data. The data can also be displayed online in the cloud with services such as Sketchfab.
LiDAR is the most common type of 3-D scanning technology. Tripod-mounted LiDAR devices like the Trimble Total Station produce detailed scans. Scanners also can be mounted on drones to survey a location in 3-D. Similarly, ground-penetrating radar is another variant, typically housed in a lawnmower-like machine that is pushed back and forth across the landscape to detect subsurface and underground objects, such as piping and archaeological features. Photogrammetry, creating point cloud data from 360-degree panoramic photos captured by drones, is an easily accessible technology, generating detailed high-resolution models.
Emily Schlickman, Associate ASLA, a co-lead of SWA Group’s XL Lab, the firm’s in-house think tank (see “Thinking Ahead,” LAM, December), says 3-D scanning is high on the firm’s list of emerging technologies to explore in landscape architecture practice, alongside virtual reality (VR) and drone-based photography. Schlickman has been experimenting with smartphone scanning apps like 123D Catch and has obtained LiDAR-generated imagery for a project site in Pacifica, California, where the firm is designing a mixed-use development on 86 acres of coastal property.
One of the more tantalizing possibilities is combining 3-D scanning imagery with VR as a next-level visual communication tool. “We just walked a client through a design in VR for the first time,” Schlickman says. “I can easily imagine sending a Google Cardboard to a client with a link to eight or 10 panoramas of a project to give them a better understanding of a site.”
Three-dimensional scanning has obvious applications for landscape analysis, as well. Rather than rely on government-generated GIS data or conventional aerial photography—which is very expensive if you wanted to, say, shoot a particular area on a monthly or even yearly basis—drone-based photogrammetry might allow firms to create their own custom imagery on demand. For example, Jones uses a Phantom 4 drone outfitted with a GoPro camera, which is accurate to three to six inches, for photogrammetry work. With that setup, he can cover about 60 acres in an hour.
Anya Domlesky, Associate ASLA, the other co-lead of XL Lab, used LiDAR imagery in her research on coastal erosion while a student at Harvard’s Graduate School of Design, and envisions 3-D scanning as a primary tool for landscape architects to understand and communicate the impacts of sea-level rise and climate change and to design interventions.
“It’s useful to understand dynamic natural processes, such as how sand dunes move over time and how engineered systems like coastal revetments influence wave and wind action,” she says, noting that two SWA Group staff members recently obtained commercial drone licenses, putting the power of photogrammetry within reach. “It’s something that’s on the horizon for us. You can really see a lot of promise with this technology at a regional system scale.”
But is current 3-D scanning technology applicable at the site-specific scale where landscape architects primarily work? Not so much, Domlesky says. “On the site-specific side of things, it may be three, four, 10 years before it’s applicable.”
Point cloud data may be uploaded into software packages such as ArcGIS and Autodesk that landscape architects are accustomed to using, but that doesn’t mean the resulting files are very practical. Donnie Longenecker, ASLA, a lecturer in the landscape architecture program at the University of Georgia (UGA) who has consulted with Jones’s company, LandAir Surveying, on how to make 3-D scanning work for landscape architects, says the problem is that currently available software for processing point cloud data is geared toward architecture and engineering. Roads, buildings, and other rectilinear objects are easier for the software to filter out of background “noise”—not so with the organic forms of the landscape.
“With buildings, it’s pretty simple, because you have walls and roofs, and [the software] can select that,” Longenecker says. “But when you have an uneven ground plane, trees, plant material—it all gets muddled together. There is no processing program to cull the data down into sets of information that can be easily turned on and off when it comes to site-specific stuff.”
Jones provides such processing services for clients, and also offers support services for clients who want to wade into the technology on their own. It’s a matter of paring down the point cloud to what you actually need, he says. Because point cloud data is so dense, he recommends starting by instructing the software to delete three out of every four points, which still provides ample resolution for landscape applications. And, because scanners take in everything in sight, it’s important to lop off whatever might not be needed in the distance or foreground of the image, a crucial step that results in the ghostly, oddly attractive black background images most people often associate with 3-D scanning.
It may be a while before 3-D scanning becomes part of landscape architects’ daily work flow, but its peripheral applications are already proving valuable. Cari Goetcheus, a landscape architect who teaches in UGA’s historic preservation program, is using the technology to document cultural landscapes at Virginia’s Stratford Hall and at Wormsloe, an 18th-century plantation on the Georgia coast.
Through a grant from the National Park Service, Goetcheus and her collaborators at UGA’s Center for Geospatial Research are currently evaluating different forms of 3-D scanning for documenting historic sites, with hope of defining which contexts merit a professional surveyor with expensive LiDAR equipment versus a landscape architect going out with a low-end drone or just using an app to process scans from a smartphone.
Her team is also trying to assess whether 3-D scanning is even preferable in certain instances. Might an old-fashioned photograph suffice at times? Does cumbersome technology not get in the way when all you want is to know whether the forest is dominated by pines or by oaks?
“These technologies can be either cost-prohibitive or knowledge-prohibitive, versus accessible and reasonably priced,” says Goetcheus, whose grant project will result in a guide for using 3-D scanning in cultural landscape inventories. “Sometimes you want to know all the topography and hydrology and ecological systems, and sometimes you just want to know what kind of metal that gate latch is made of, or what historic pattern that fence was built in. We’re trying to figure out the differences between all the different types [of 3-D scanning] for people who aren’t so tech-savvy, and then for people who are really savvy.”
The data sets that result from 3-D scanning are enormous, and can crash a computer in a heartbeat. It’s not unusual for LiDAR data to take days, even weeks, for software to process, which is why Goetcheus is fond of programs such as Agisoft PhotoScan, which she considers do-it-yourself-level software. “It’s almost as easy as taking your pictures, uploading them to the software, leaving your computer on for four hours while it is crunching the data, and then it is going to make you a three-dimensional point cloud.”
Alexander Robinson, ASLA, the director of the University of Southern California’s Landscape Morphologies Lab, has experimented with 3-D scanning, but says that he, too, is wary of getting mired in the technology before it’s been refined for the purposes of landscape architecture. “We’re still figuring out how it can engage the skill sets of landscape architecture,” he says. “People want to go out and scan something, but what is it going to add up to? It’s really fun technology, and it’s very seductive that way.”
At the very least, 3-D scanning provides information about the environment that was not previously available. What was once the exclusive purview of the military and the world’s biggest technology companies is now in the hands of the average design firm—a powerful tool with as yet unimagined applications for the common good. Despite his reservations, Robinson speaks of the possibilities in the grandest terms: “Just like the invention of the camera, this is going to change things.”
Brian Barth is a Toronto-based writer focused on culture, design, energy, and the environment.