This lecture was presented at the 3D Digital Documentation Summit held July 10-12, 2012 at the Presidio, San Fransisco, CA.
Documentation to the Secretary of the Interior’s Standards: Assessing the Value of Laser Scan Data
The Heritage Documentation Programs consider High-Definition Surveying (HDS) to be a significant tool, one of many it employs in the survey of historic sites and structures. However, this technology by itself is limited in its ability to provide adequate information to completely document heritage sites to the high standards recognized today by the preservation community. HDP has utilized terrestrial laser scanning in documenting cultural heritage through experimentation and application of the technology since 2002 and continues to incorporate its use extensively into their workflow. With a mission that places emphasis on creating an archival record, HDP strives to supply project sponsors with a comprehensive set of deliverables that convey an understanding of a site or structure to the general public; interpret its processes, patterns of use, and cultural values, and provide baseline documentation for rehabilitation and restoration. Research and data capture necessary to fully describe historic architectural resources requires an understanding of the principles and history of architecture to help define, manage, and guide the documentation effort. The trained staff of architectural historians, architects, landscape architects, and engineers at HDP provides a discerning eye to projects to make informed decisions from laser scan and field data that ensures knowledgeable and sound documentation. This multi-disciplinary expertise is also utilized in the HDP summer intern program to mentor and educate the next generation of architectural preservationists, providing longevity to the many techniques and methodologies of documenting our cultural heritage.
As laser scanning greatly reduces the time needed in the field for measuring, it also tends to reduce physical contact and exploration of a site that can uncover or expose unexpected features not readily seen. While it remains virtually impossible to capture 100% of a site or structure with laser scanning alone, combining HDS with other measuring techniques and extensive research has proved an effective means of gathering field measurements and data that reinforce thorough documentation. In addition to supporting large-format photographs and historical reports, the creation of standardized, conventional drawings facilitates strict archival stability standards and the public and scholarly dissemination of the documentation. HDP uses supplemental data to fill in the blanks left by laser scanning, allowing for the reverse engineering of point clouds into smarter parametric 2D drawings, 3D solid models, meshes, and surfaces. These models and other visuals created during the project workflow can be manipulated to provide a multitude of products determined by the sponsor’s needs. The printed reports, photographs, and drawings become the archival material that will secure exceptional permanence for the documentation while the digital data remain at this time unconventional and unverifiable entities. The Library of Congress and others are researching methods and formats in which to sustain “born digital” records to standards defined for the collection, but at this time none have been propped.
The emergence of new digital HDS technologies has increased the ability to measure heritage sites faster than ever before, but a hasty application of these technologies can easily result in superficial and incomplete documentation of the significant features of a structure or site. To achieve well-examined, thoughtful, and comprehensive documentation, laser scanning must be supplemented with additional field measurements and observations and receive specific evaluation and translation by professionals in the field of historic preservation.
Cordell: … definition survey, high dynamic range pano-photography, and CAD to produce archival drawings of historic structures and his most recent projects include Ellis Island, the Ghazni Towers in Afghanistan, the USS Cairo, and Prince William Forest Park in Virginia. So we welcome both of them this morning. Thanks.
Lockett: Good Morning. I’ll go through this really quickly because we’re very limited on time. Paul and I are here to talk about the Secretary of Interior Standards and how the heritage documentation programs are integrating laser scanning into its work flows to meet those standards.
I’ll start by giving you a little background on the programs as well as the preservation laws that govern our work, then I’ll show you some specific examples of how the Historic American Engineering Record uses laser scanned data. Before turning it over to Paul, I want to talk about some of the work that the Historic American Building Survey is doing.
The Historic American Building Survey was established in 1933 as part of FDR’s New Deal Administration with an agreement between the National Park Service, the Library of Congress, and the American Institute of Architects. The Historic American Engineering record and the Historic American Landscape Survey were established later but all these programs always adhered to the initial mandate of documenting America’s historical cultural resources on a national scale with uniform standards and guidelines and create a comprehensive public archive of measured drawings, historical reports, and large format photographs curated by the Library of Congress.
The National Historic Preservation Act of 1966 set firmly into Congressional legislation a federal preservation system that not only created the National Register of Historic Places and the National Historic Landmarks Program, but also reaffirmed HTP’s role in protecting historic resources. Referring to Section 106 compliance, Sections 101 and 110 of the Act legally required that first, documentation must be completed to HABS standards and second, that this documentation must be deposited into the Library of Congress prior to any federally funded action resulting in the substantial alteration or demolition of historic property.
While standards and guidelines for documentation were stated in bulletins published by the NPS architects during the early years of HABS, later amendments to the Preservation Act included a definitive set of four standards collectively known as the Secretary of Interior Standards for Architectural and Engineering Documentation. HTP has defined specific requirements for meeting these standards and we’re working on adding guidelines for the use of digital measuring devices.
Let me briefly run through the four standards as they exist currently and discuss implications for the documentation submitted to the collection. Standard one states, “Documentation shall adequately illustrate what is significant or valuable about the historic site.” Significance can vary considerably from structure to structure. For example, the USS Cairo is a resource where that ship’s structure is significant in and of itself. The Shenandoah-Dives Mill on the other hand, is significant for the processes that take place inside the structure, in this case the flotation methods of extracting metals from raw ore.
Standard two states that, “Documentation shall be prepared accurately from reliable sources to permit independent verification. This refers to drawings that are based on carefully measured archival field notes or on measurements taken digitally with the device parameters clearly noted on archival metadata forms, reports that are based on primary sources and photographs taken with large format view cameras.
Standard three states, “Documentation shall be prepared on materials that are readily reproducible, durable, and in standard sizes.” For materials testing, it was determined that drawings must be plotted on Mylar vellum, reports printed on archival bond, and photographs developed on polyester based film and printed on fiber based paper. Additionally, all of the records must be produced with carbon based black inks or colored pigments, dye based are unacceptable at the moment. For drawings, HTP and the Library of Congress have implemented three standard sheet sizes that can be easily stored and maintained in the library archive.
The last of the four standards states, “Documentation shall be clearly and concisely produced. Per our guidelines, drawings and field notes should be created with an orderly hierarchy of visual components and text. Reports should be created using the Chicago Manuel of Style and photographs should include duplicate images, one with a visible scale and one without. Also each should be framed and viewed to correct perspective distortion at the time the photo is taken. ”
Though none of our guidelines fully address digital data yet, the heritage documentation programs have come to consider terrestrial laser scanning a significant tool. Over the last several years it’s become a regular part of our tool kit we use in the survey of historic sites. But this technology is limited in its ability to fully capture and convey an understanding of the site or structure to the general public. While it may be visually compelling, a point cloud alone cannot interpret processes, patterns of use, or cultural values. Nor is it sufficient in providing baseline documentation for rehabilitation and restoration always.
When combined with other measuring techniques and in-depth research however, laser scanning has proved an effective means of gathering field measurements and data that reinforce thorough documentation. Other tools we use extensively include hand measuring, the total station, photogrametry and rectified photography, and GPS. Regardless of the methods used, our chief objective continues to be creation of measured drawings on archival materials as required by the collection at the Library of Congress.
Now let’s talk about some the sites that HAER has documented with the help of laser scanning. The program has been involved with laser scanning since we began the Statue of Liberty project in 2002. A project that took four years and some tears and learning the do’s and don’ts of laser scanning before getting it right. More recently, HAER began documenting the space shuttle Orbiter and the Discovery in 2011, and I’ll use the Discovery to discuss some of the many post-processing methods that we use to generate the 2-D drawings as well as where we needed to supplement the point cloud with information from other means. Scanned data of both the interior and exterior of the shuttle was acquired by Smart Geometrics, whose expertise with faster phase based machines proved key due to the limited time we had access to the shuttle. It’s important to note that while we observed the scanning in this case and knew what to expect in delivery, scanning by third parties unfamiliar with the subject being scanned can often deliver point clouds that are inadequate or ambiguous to the individuals interpreting the data.
Once registered and cleaned the scans were translated into smart parametric entities after extensive research and study of the shuttle’s structural elements. The interior scans were translated into 3-D solid models within autocad using cubit points sense. We found that these had to be supplemented with hand measurements, as well as data taken from existing drawings in order to model the structural members hidden within the orbiter’s skin.
The exterior scans of the shuttle were meshed using rapid form software defining lines in the skin such as the joints between the thermal protection tiles were extracted from the mesh and imported into the final 3-D models. This let us emphasize or diminish edges within the weights to maximize clarity of the final 2-D drawings. The meshes from rapid form and solids from autocad were combined inside Rhino and then exported as 2-D poly lines for drawing assembly back in autocad. This let us generate standard plans you see here, sorry about the quality of that, and also isometric exploded views. Modeling the shuttle in 3-D really greatly increased our versatility and the speed in producing the 2-D drawings that are clear and concise. But there are several significant systems and processes associated with the Orbiter, processes that should be illustrated as part of the engineering record. In this drawing, careful research allowed us to illustrate the shuttle’s reentry into earth’s atmosphere and it’s landing sequence. These processes would most likely be overlooked without diligent investigation into the shuttle’s many functions.
And now I’ll turn it over to Paul.
Paul Davidson: So, I’m going to talk about documentation of historic buildings using the mess hall at James. H. Quillan VAMC as an example of how we incorporate laser scanning into our work flow and supplement its weaknesses with other methods of field recording.
We began with a ten position traverse of the exterior and this quickly let us define the exterior floor plan, locating most of the window and door openings and giving us most of the dimensions we needed for the elevations. We did one interior scan and the large open space originally served as the dining hall, and now it’s used by a museum of medical history, and we chose to laser scan this space because all the artifacts would have made it difficult to measure by hand. The scan allowed us to quickly layout the basic geometry and locate doors, windows, columns, and overhead beams. The rest of the plan consisted of more than fifty separate rooms and numerous closets which we field sketched and measured by hand. Laser scanning these spaces to generate a floor plan would have been far too difficult and time consuming. For one, many rooms would have required more than one station with most having obscure lines of sight due to furniture and storage and two; it would have been difficult to create the controlled network we’d need to link up those scans with the exterior scans.
Even so, the exterior scans provided a highly accurate rigid outer envelope by which we were able to judge the accuracy of the interior dimensions we took by hand. But these dimensions alone do not complete a floor plan. There are details too. This leads me to what we feel is the biggest difficulty with laser scanning, which is its limited ability to capture molding profiles. Of course, it’s true that simple large scale profiles with exceptional scan coverage can generate a useable profile, like this bracket under the eaves of the clock tower. But smaller scale geometry isn’t as easily discernible like this crown molding in the same tower.
When it comes to window and door profiles, the results are even less readable. The window from the west elevation is well defined at a resolution of about two millimeters. The close up on the right shows that the scan does contain some good detail, at least as far as things like the sill, the arch, the brick coursing and bond are concerned, but the real problems arise when you attempt to extract information from the heavily painted sash and jam. So there are two issues here and they’re clarity and efficiency. In plan, you can see that the scan lacks the clarity needed for extraction. This is for a variety of factors, including the small shapes involved, such a s beads, moutons and stops as well as a significant build up of paint. Noise, edge effects and shadows in the point cloud also complicate analysis.
Of course, further processing or denser scans might allow for a clearer result but wouldn’t allow for layers of paint which fill in small beads and soften crisp edges. When a profile is taken using a comb on the other hand, an experienced individual can account for this build up effectively reading the profile beneath the layers. Frequently the difference between a simple and compound curve is felt, it can only be felt, so it’s something too subtle to determine in a fuzzy line of points.
In terms of efficiency, the scanners weakness lies in its inability to capture an entire window profile from a single position. A complete scan of a window jam and sash would require multiple positions with the sash both opened and closed and it would still probably miss how the meeting rail joins. There’s simply no justification for taking the time and effort to do this when a higher quality detail can be achieved in a few minutes with a profile comb and a measuring tape. The issue is more apparent when you see the CAD drawing of all the different window and door profiles taken at the mess hall. Laser scanning all seventy doors and all sixteen window types at an appropriate resolution would make for a monumental task. Those tend to be washed out.
The finished product shows the level of detail and clarity required for a well rounded set of HABS drawings and the profiles of this project were relatively simple compared with some other projects, so I wouldn’t want to try to derive some of these profiles from a laser scan. And then we have details that would be, this is a window detail, but it would be far more difficult to do this one with the embrasured shutter details so you would have to scan it opened and closed in some manner. In this detail at Prince William Forest Park where the sill flips open and the sash is stored below, it’s obvious the laser scanner isn’t the best tool to document this. So we’ve heard a lot about sculpture but we have some problems with laser scanning sculpture. Though the laser scanner has the advantage in representing this three dimensionally, when it comes to generating a 2-D line drawing, we found rectified photography to be more useful. As you zoom into the point cloud, it becomes more and more difficult to discern the hard traceable edges. A mesh of the area might clarify this but it could also soften the finer details of the sculpture. In cases like these, the laser scanner is more useful as a reference to judge the accuracy and location of the finished product and this is, you know, like thirty feet in the air, so we’re capturing this detail from the ground. That’s a very detailed 2-D drawing. Visualize it.
So we know that there are scanners capable of producing a higher level of detail but weighing the expense of the equipment against our end product pictured here, we’ll probably continue to use photographs for some time to come. This drawing, though greatly reduced in resolution, the original is 34′ by 44′, begins to show you how a line drawing produced by a skilled draftsman creates a hierarchy that highlights the more significant features of the building, something that’s hard to obtain in a screen capture of a point cloud.
So I’m going to talk a little bit about high dynamic range pano-photo texture mapping and the problems we’ve had with that. Initially, we thought this would help us clarify our laser scans, although we found that photo texturing doesn’t increase our ability to extract the information we need from the point cloud and we found that the best information we can get is from a finely scaled intensity map. The biggest stumbling block is the disparity between image resolution and scan resolution. What you see on the screen is a texture map of the interior of Castle Pinckney in Charleston Harbor created using a typical set up of a DSRL camera equipped with a fish eye lens mounted on a Nodel bracket . The scanner is positioned about twenty-five feet from the wall and in this instance the resolution of the pano-photo matches the resolution of the point cloud and the resolve is amazing. However, this station is a much greater distance from the fort at sixty-five feet from the scanner, the photo has only six hundred thousand pixels to map to 3.5 million data points. This means that the results are interpolated through sub-sampling resulting in a less sharp texture map and you can see the brick joints are more difficult to define. The problem becomes even more pronounced at the hundred and seventy foot mark. Of course this can be mitigated using lenses with longer focal lengths or positioning the scanner closer to the subject, thereby limiting the range of the scanner. But correctly determining distance, focal length and scanner resolution may be difficult and time consuming especially over larger or more complex sites.
Even with good resolution, results can be lackluster in areas, the presence of chromatic aberration or color friging is particularly vexing. These effects can be reduced in post processing but this adds extra steps to an already time consuming process plus it cannot be dismissed that HDR photography is ultimately an artificial portrayal of the color of an object and added post processing only increases the contrived nature of the image. Even with good mapping results, meaning low pixel or color information can be mismatched. That is particularly a problem on thin elements like the rigging of the USS Cairo or branches against the bright sky. And it also occurs when the subject moves between the time it was scanned and the time it was photographed, introducing noticeable mapping errors in things like trees and grass. With these drawbacks in mind, ACR texture mapping technology does produce useful tools that can aid in interpretation. These views, essentially photographic sections, would be impossible to capture with a camera onsite. Imagine an awesome fly through. So that another product you could create via fly through of this and you can go into spaces that the public wouldn’t normally be able to access.
So clearly over the course of our experience using the laser scanner, we found definite advantages and disadvantages with the technology. Some of the advantages it offers in the field include the potential to reduce time spent measuring large non orthographic sites and structures, its ability to capture enormous amounts of measured data, and that it can be used as a non-contact technique for sites where there are contamination or instability issues. In the office, using laser scan data has made it easier to model sites in different formats though 2-D drawings continue to be our main output, the by -products of post processing clean point clouds, meshes, and solids, allow us to offer a wider range of 3-D products that can be used as interpretive visuals by parks or project sponsors such as virtual tours, fly throughs and animations.
Some of the disadvantages include difficulty in obtaining one hundred percent coverage, difficulty in obtaining accurate information for small scale details, the high cost of purchasing hardware, multiple software packages, subscription services and the training needed to use it all can be prohibitively expensive. It can limit exploration of details, the time savings in the field is often offset by longer post processing times in the office and lastly, one of the biggest shortcomings is that born digital data produced by the technology does not meet the archival standards that our collection must adhere to.
So in conclusion, we strongly believe that in order to achieve comprehensive documentation, laser scanning must be supplemented with additional field measurements and observations and need evaluation and translation by professionals in the field of historic preservation.
Dana Lockett is an architectural project manager with the National Park Service’s Historic American Engineering Record in Washington, D.C. He holds a Bachelor of Architecture from Texas Tech University, and has 22 years of documentation experience with Heritage Documentation Programs. While still a fan of hand drafting, Dana emphasizes digital documentation of engineering and industrial sites using high-tech measuring devices such as High Definition Surveying (HDS) combined with 3D reverse-engineering software and CAD. His most recent projects and collaborations include the Statue of Liberty, Grand Canyon’s Grand Gulch Copper Mine, NASA’s Rocket Engine Test Stands, and Hawaii’s Pu’ ukohola Heiau NHS.
Paul Davidson is a Project Architect with the National Park Service’s Historic American Buildings Survey (HABS) in Washington, D.C. He holds a Bachelor of Architecture and Certificate of Preservation from Pratt Institute, and has 10 years of experience with the Heritage Documentation Programs (HDP). His projects at HDP combine hand-measuring, High Definition Survey, High Dynamic Range pano-photography, and CAD to produce archival drawings of historic structures. His most recent projects include Ellis Island, Ghazni Towers of Afghanistan, U.S.S. Cairo, and Prince William Forest Park.