This presentation is part of the 2017 3D Digital Documentation Summit.
Incorporating 3D Digital Data in HABS/HAER/HALS Projects
Architects Dana Lockett and Paul Davidson of the HDP presented four years ago on our use of High Definition Surveying to gather field data for cultural resource documentation. In combination with other techniques, both traditional and digital, laser scanning continues to be a key tool in nearly every project HDP undertakes. While our stable of project deliverables has expanded to include digital offerings such as hi-res renderings, fly-throughs, mesh models, and virtual tours, the production of measured drawings continues to guide our use of the technology. Our mandate to create documentation meeting both HDP and the Secretary of the Interior’s Standards leads us to examine in greater detail the accuracy and precision of our data collection methods. Issues with scan data that may be unnoticeable or overlooked when creating renders or animations cannot be ignored when creating detailed line drawings. In this year’s presentation we will discuss practical issues encountered in the field, and some of the resulting approaches we have adopted for the different kinds of resources we document, illustrated with recent projects.
The massive jump in speed between recent iterations of scanners means that we are now utilizing laser scanning in scenarios we would not have previously contemplated. Larger and more complex sites are becoming easier to document, the rapid scanning of building interiors is now feasible, and both can be done at higher point densities than ever before. The ability to accurately register interior and exterior data to create a more complete model of a building has proved invaluable, particularly with historic structures, whose surfaces are rarely plumb or level. No longer do we need to create networks of triangulated measurements and diagonals just to determine the squareness of a structure in a single plane.
But combining many dozens of stations is not without challenges, regardless of whether target-based or cloud-to-cloud registration is used. In either case, minor changes in the way the software is registering the clouds and averaging the error between them can cause alignment issues that are not easily detectable at first glance. Unfortunately, No matter the registration method a close visual inspection is the final step in determining the accuracy of the registration, even when weighed against total station control networks. At the moment we prefer target-based scanning. Because we feel the results from cloud-to-cloud registration can be ambiguous, recently we have been surveying the target network with a total station and using those points as a framework for assembling the scan data.
Our experience with scanning has also led us to conclude that more points do not necessarily mean more detail. Data collected at Ellis Island and Kalaupapa National Historic Site show that scanners still lack the fine grain and surface precision required to create drawings of millwork. While current machines do an admirable job with filtering, there is enough “noise” that things such as door and window frames, crown moldings, and muntins with 1/8- to 1/16-inch elements are difficult to interpret. Accurate drafting of these elements is critical in our work: our drawings are often used for facilities maintenance, repair or reproduction, or may be the final record of an historic resource facing demolition. Traditional hand-measuring with profile combs and tapes continues to prevail in these instances.
Likewise, digital photogrammetry is our preferred method for documenting most sculptural ornamentation, inscribed text, and free-standing statuary. In addition to producing excellent scaled mesh models, photogrammetry generates high-resolution ortho-images, which have much better rendering of fine detail than a photo-textured point cloud and are better suited to the production of drawings. We recently used photogrammetry to capture the many sculptural elements at both the Meuse-Argonne and Aisne-Marne American Cemeteries in France, including cast-bronze flagpole bases and individually carved column capitals. This technology has also proven useful with capturing the finer detail required for several projects done the National Capital Region, where inscribed text has been damaged, such as on the African American Civil War Memorial and Korean War Veterans Memorial.
Obtaining accurate data beyond the range of the machine and with non-reflective surfaces remains problematic. As a result, we are exploring alternatives for resources that are, for example, extremely tall or darkly colored. Both issues were present when we attempted to document the Cape Hatteras Light in North Carolina. Not only is it the tallest lighthouse in the United States, standing at 210 feet, but it is painted with a black spiral stripe that terminates in elaborate cast-iron brackets supporting a two-tiered catwalk and lantern. After initial attempts to capture the top of the lighthouse with our machines proved unsuccessful, we found a solution in combining “short-range” scans from our data, “long-range” scans from a Reigl machine, and photogrammetry taken (somewhat precariously) from the catwalk levels. With a 2017 project planned for Cape Lookout Light, we will continue to experiment with combining different kinds of data to overcome these limitations.
Lastly, there are some issues for which we have yet to find effective solutions:
Larger scanning projects and faster collection of points also means larger data sets. In the last two years, HDP has undertaken a number of projects that document large landscapes and structures, including the two previously mentioned WWI American Cemeteries in France and Fort Jefferson in the Dry Tortugas. These data sets are becoming so large that they are exceeding the ability of software programs to manipulate them without freezing or crashing, despite ‘ performance claims. Inspecting, animating, and drafting from point clouds have become more time-consuming and laborious as a result. Transfer of large files to project partners and long-term storage of digital data also remain problematic.
Limited time on site means we have been forced to scan in less-than-optimal conditions. While rain causes problems with imaging, it is wind and surface stability that will decide whether scan data is usable. Experience at Fort Jefferson and Kalaupapa National Historic Site has taught us what wind speeds—and what kinds of winds–will still produce usable results acquired within the tolerance of the machine’s real-time compensator.
The mission of the Heritage Documentation Programs has remained virtually unchanged for the more than 80 years it has been in existence, but our mandate to produce documentation of the highest quality has, paradoxically, often put us at the forefront of technological experimentation. Each project we undertake has its own interesting challenges, and each provides an opportunity to assess our practices and experiment with new ideas. Laser scanning still has not supplanted the old methods in some areas, and in others we have found newer alternatives that provide better results, but it continues to be an important tool for the HDP. With scanning’s continued importance we remain keenly interested in ensuring that we collect data with the highest accuracy possible.
Daniel De Sousa is an architect with the National Park Service’s Historic American Buildings Survey in Washington, DC. He holds a Bachelor of Arts in Architectural Studies from Connecticut College, and has nine years of documentation experience with the Heritage Documentation Programs. Though skilled with profile comb and pencil, he has also completed the comprehensive scanning of several large sites and structures, and instructs interns in the HDP’s annual Summer Program on creating architectural drawings from point cloud data. His most recent projects include Ellis Island, Cape Hatteras Light, Fort Jefferson at Dry Tortugas National Park, and Kalaupapa National Historic Site.
Ryan Pierce is an architect with the National Park Service’s Historic American Landscapes Survey in Washington, DC. He holds a Bachelor of Arts in Art History from Wake Forest University, and a Master’s of Science in Historic Preservation from Clemson University and the College of Charleston. He is continually honing his laser scanning and photogrammetry skills, and working to expand the reach of digital products that result from current workflows into usable interpretive tools. His most recent projects include Aisne-Marne American Cemetery and Memorial, Cape Hatteras Light and quarters, hangars and gun emplacements at Gateway National Recreation Area, and Kalaupapa National Historic Site.