This lecture was presented at the 3D Digital Documentation Summit held July 10-12, 2012 at the Presidio, San Francisco, CA
Application of Current 3D and Pseudo-3D Imaging for Conservation
Melvin Wachowiak, Senior Conservator
E. Keats Webb, Digital Imaging Specialist
Smithsonian Museum Conservation Institute, Washington D.C.
Conservation professionals have been aware of modern methods of 3D digitization, but a comprehensive guide including the strengths and weaknesses of available technologies is not available. A reference that includes differing results, and addressing the most appropriate fields of application will help these professionals incorporate 3D and pseudo-3D imaging. With goals of broadened access, preservation, supporting education, and for enriching context, cultural heritage professionals are turning to digitization for the documentation of museum objects. Digital imaging of collections includes capturing 3D information, which supports research, preservation, documentation, etc. Digital surrogates provide easier and better access for researchers: a digital file can travel the world without the object being handled and exposed to the elements, thereby aiding in the preservation of the object. Digitization provides interactive and accessible exhibits, displays, websites, etc., and opportunities for education initiatives and means of enriching the context. Digital imaging, including 3D and pseudo-3D, can be a truly democratic, non-destructive, non-contact method of researching and conserving an object.
There are already many techniques that capture 3D information, both true 3D and pseudo-3D. 3D imaging involves surface and volume-based three-dimensional measurement that can result in physical or virtual replication. In contrast, pseudo-3D imaging techniques provide surface information but may lack in the precision or surface measurement capabilities. The imaging techniques we will be presenting include 3D scanning using laser or projected white light, photogrammetry, Reflectance Transformation Imaging (RTI), Quick Time Virtual Reality (QTVR), and HD Video. We will provide a description of these methods in the presentation along with offering guidance for the most appropriate applications. In choosing the appropriate imaging method, one needs to assess the research questions, desired outcome, and the project budget in light of the available technologies and their associated strengths and weaknesses. Understanding what is being attempted, what is achievable, and the true cost will help to inform the selection of the appropriate technique.
This presentation will follow the digital acquisition, data processing, and examine the final product. We will use the methods listed above for a single object: a carved and painted wooden sculpture. By digitizing the same object with a variety of techniques, we will be able to more effectively present the comparison of these imaging methods, including specific examples of strengths and weaknesses and other applications. In comparing the techniques, we will look at “3D-ness”, resolution, color accuracy, portability of the technique, accessibility of the resulting product, universality of delivery, skill level required, equipment cost, and archival standards.
There is no single technique that will be ideal for all objects requiring excellent spatial resolution, calibrated color, in an easy to distribute package. Each technique has its advantages and weaknesses, produces different results, and has different fields of application. Sometimes a combination of techniques is most effective. This presentation will look at realistic options for imaging cultural heritage objects along with a comparison of those techniques.
Church: We have two speakers, Melvin Wachowiak, who is a senior conservator at the Conservation Institute Smithsonian in Washington D.C. Melvin received an M.S. from Winterthur Museum Art Conservation program, University of Delaware, and a B.S. in Arts from Springfield College. He has been at the Smithsonian since 1989 and is involved in projects in the Fine Arts, Anthropology, Archaeology, and the History of Technology Collections. Since 2004, he has led MCI’s use of new and advanced spatial and spectral imaging techniques such as 3D scanning, 3D microscopes, and other computational and digital imaging. E. Keats Webb, our second speaker, is an imaging specialist in the Museum Conversation Institute, Smithsonian Institution, Washington D.C. Web graduated in 2007 at University of North Carolina at Chapel Hill with a B.F.A in Photography. In 2010 she started her own imaging business focusing on scientific and computational imaging and event photography. Her work includes scientific computational imaging research, technique development, digital assessment management, and video. Now I will turn it over.
Wachowiak: Thank you Jason. Good afternoon everyone. I am from a branch of the entertainment industry at the Smithsonian Institution. Keats and I work at the Museum Conservation Institute. Maybe you do know a little bit about the Smithsonian. It has a fairly large collection; 144 million objects and specimens, 19 museums and galleries, the National Zoo, a number of research centers, more than 12 thousand employees and volunteers. Twenty four billion visitors annually, 183 billion web visitors a year. This year we are on track to have even higher visitorship in both categories. We are pilgrimage destination, as we say. A lot of people come and don’t know why they are coming, but there is a whole mess of them.
As I said we’re at MCI, and MCI is a smaller group among the Smithsonian. There is about 25 full time employees and at anytime we might have ten or more post-docs and other fellows. We have lots of interns this summer, so it’s pretty wide ranging group that includes not only the imaging studio that Keats and I work in, but a number of material scientists, chemists, doing analytical chemistry, which includes protein chemistry, x-ray diffraction, Raman spectroscopy, SEM work, and on and on and on, just to give you the scope of our work, we’re just one small part of the whole enterprise but we work together on projects. Imaging is an intrinsic part of many, many of the projects. We also do multi-spectral and hyper-spectral imaging, infrared UV imaging, gigapixel imaging, 3D microscopy, but again, this is an important part of what we are doing.
The subject of today’s talk, looking into different methods for collecting 3D and pseudo-3D information. What we’re going to do is look at an object, a humble carved wooden object which is painted and it’s a small object, it’s not a building, we don’t do architectural things, typically. It would be unusual for us to be involved in large architectural projects. And what we wanted to do was get a sense how a number of these techniques: RTI, QuickTime, Virtual Reality, 3D scanning, photogrammetry, and how they fit in with our workflow, what parts of it translate to the shop floor.
I am a conservator, I’m not a scientist; I don’t even play one on TV. What I do is support other researchers and our own research across the institution and occasional collaborative work outside the Smithsonian. I already have my profession, I don’t need a new trade, but there is plenty of ways for us to cross-collaborate. So what we are going to look at is a number of these techniques and our nacient experiment, to look at the same object with these different techniques and what we’ve compiled in your handout is, I think, a better way to present what we ended up with and to try to put it in a PowerPoint. You can read it as you have time later. I will go over some of the most important aspects and you can view it at your leisure. Probably the easiest way for us to do this is we’ll run through a short video, the techniques I am going to show you and the application side of it. Again, this is a small wooden object, maybe a foot across. It’s carved out of soft wood. The underside is not coated with paint, it’s somewhat reflective but it didn’t really pose any problems for any of the techniques, or it shouldn’t have because it’s not too glossy, it’s not too dark, it’s not too bright, it is not exceptional.
A quick run through the techniques; the QTBR is first. We used 36 images set up on a stage. We tried to light very equally and pretty much the same idea here with the HD video, it’s a very quick technique as Keats will allude to it; it doesn’t surprise you that you would conceive of this as a 3D object. RTI has been brought up a number of times so luckily, you don’t have to hear us tell you about it again. We will go through the product of the RTI but nothing in the conceptual methods. Photogrammetry has been covered a lot too, we’re lucky we are at the end of the day. Probably none of this is surprising to you. This was using a 1 2 3D Catch to make a 3D model. We have a structured light scanner, a Breuckmann structured light scanner. There are a lot of products from that. Of course you have the points, the mesh, you have substandard color capture, but you do have color.
The first area that we produced the sort of baseline of a very high quality data set was 3D scanning. Our Breuckmann scanner for this project, was sort of at a mid-level resolution, which is about 60 micron, which is a lot different than architectural stuff. It is pretty rapid. The amount of data you get is quite large but the file size is not huge, and I am talking about a couple hundred of megabytes, it is not a tremendous amount of data. This is a true 3D capture technique, very high resolution. You get good metrology especially on a subject like this. You have a lot of potential to reprocess and do comparative analysis in a large group of objects and many products can come from that, both virtual and real. You can go back to reality and print or mill. The downside of it is that you cannot scan highly reflective or translucent surfaces or black or white objects or black and white together and you need a low ambient light. So it’s not good for high noon out in a field but with control of ambient light, you can certainly do it offsite with a battery or a generator. The equipment costs are pretty high comparatively, again in a comparative way. You do need engineering skills to master. The color capture is fair to middling and that said, it’s still metrologically the touchstone.
The other technique we used was photogrametry, 123 Catch, it’s simple. The processing is done in the cloud, so it is pretty low in terms of computational need. This too is a true 3D product and just for the sake of time, I’m going to move a little bit more quickly. Like I said, you can certainly read this yourself. We had good results but not always, so it’s kind of surprising when it would work and not. I wanted to point out that the way we set up was using a stage and rotating the object and having really tight control on the lighting. So if something went wrong we knew it was the object or the processing and not our setup.
Webb: So, jumping into the pseudo-3D imaging, one of the things we would like to talk about with the imaging studio and the imaging we do there, is we have 2 ½ DM pseudo imaging techniques. So, they’re not giving you a mesh or a point cloud or the measuring capabilities of a true 3D imaging process. But in some cases, pseudo-3D imaging is enough, so this is jumping into that pseudo-3D imaging, reflectance transformation imaging.
The top image is a composite image that just looks really cool. So it’s strips of the specular enhancement without the color component, and then also just the normal color image, and again, it just looks really cool. This is our setup, that’s unique to our studio right now. RTI is great because its low equipment costs, a lot of the stuff Carla covered. A lot of people have this equipment already so it’s not a big jump in buying a new scanner. You have a high resolution based on the camera system that you’re using. The final product is interactive and it has a simple user interface where you can continue your research and use that. It’s portable, a pro and a con is that it is still being developed. That can be a really good thing but it also means that there is something that we are still waiting on, including the measurement capabilities and its open source. So there are challenges with RTI, it’s not perfect on everything, but it’s also a great candidate for some objects.
QuickTime Virtual Reality is very much a pseudo-3D imaging technique. It gives you the round view of an object. You can see all the way around it and in this case, we did the top part of the mask and also the underside. So if you just need to see the object and not make any measurements, it can be a great technique. We used a Lazy Susan, a turntable, did 10 degrees, and ended up using the same setup for photogrametry also using the same image set. So it’s low equipment costs, moderate skill level, accessible product, you can either have it stitched together so you can move the object around yourself, or it can be done as a video and it just moves on its own, and again, no measurement capabilities and it lacks that true 3D information.
The last technique is high definition video, very similar with QTVR. We used a turntable, turned the object so you could see the underside and the topside. Fairly low equipment costs, a lot of DSLR’s now have HD video capabilities, so that can be very easy, it can be very accessible on YouTube. A lot of people have YouTube on their phones, their tablets, so they can see that, and if that’s the needs for your project then it can be really great. But again, it’s lacking that true 3D information, and it doesn’t have measurement capabilities. You are also throwing out some of that resolution with the video.
Wachowiak: I forgot two things; one was that our new secretary, the head of the Smithsonian has said publicly and in a number of interviews since he arrived, that he wants to digitize the entire Smithsonian collection. Now we are not going to do it all, not 144 million but it’s going to be at least 10%. So we are going to have to get cracking I think. One of the things that we like about the 3D scanning and the photogrametry is we could pass through that data and generate Adobe Reader accessible images that anybody could do cut throughs and take measurements, as long as your metrology is included in your model. We’ve satisfied one of our needs and that is for the broadest dissemination. One of the problems that we’ve had over the years is if you’re working with a curator or scientist and you say, “Yeah, here’s the DVD with your data” and they ask, “How do I look at it?” You say, “Well you need a 10 or 12 thousand dollar licensed engineering software and you can do whatever you want with it.” That is not really helpful. So what we strive for is to give them as much as possible, and something as accessible and potentially long lived as a PDF archive document could be a real help for us.
This is a synopsis of what the chart is that you’ve been given. What we’re really concerned about is the truthiness, the 3Dness of anything, and obviously, with 3D scanning and photogrametry you have the data that you can see for yourself. When you have digital video, Quick Time Virtual Reality, it’s easy to convince a user that they are seeing a 3D object, they have already been fooled their whole lives, they’ve seen movies, you don’t have to explain, the refresh rate is so fast, you can’t even process it. It’s not really 3D. They just assume that it is, so if it is about sharing for the widest possible number of people, then those techniques are really good.
With things like resolution, the 3D scanner is top hands down, that’s not even a question, but where we might have had millions of points for the scan data, with the photogrametry, the 1,2,3D Catch, we had over fifty thousand points, that was pretty good data. If you are doing comparative analysis, even a visual analysis, there is a lot of data in that mesh. If you are doing things like planning exhibits, storage, building packing cases, there’s probably ample data in that. So depending on your purpose this is probably a worthwhile investment, especially in the short term. We have no illusions that we will be repeating these exercises. We are not going to start from one end of the Smithsonian with the equipment and the techniques that we have and come out next year having finished everything, and that will be it, now we can go on to other things. It’s going to be a long term process, but it’s obvious, I don’t have to convince you that the third dimension exists and is part of our daily procedures.
I think I will go right to the summary remarks and we’ve probably seen enough of this over the last couple of days, it’s self-evident, but no single technique satisfies all the needs of a collection, and a third point, a combination, we may use more than one technique. We’ve seen a couple of presentations where at least two techniques were used very successfully. For us, a lot of what we’re doing is trying to get enough knowledge and experience so that we can have an intelligent conversation with an expert in the subject. So when we are dealing with the engineers we can engage in a conversation about what our goals are and what we want to get out of it and what they can and cannot do with the object. So at least we’ll pass as intelligent people.
Finally, we want to be able to help our colleagues in the collections so that we can better understand what they are trying to do and tell them when they actually arrive at a question, “What is the purpose?” We get people who come and say, I want to do RTI on this and that, well that’s not really what you are asking, or the same thing about 3D scanning. No you don’t really want to do 3D scanning. We can accomplish a lot with RTI. So we are informed and better able to inform. With that, hey I just hit zero, spot me on that.
Melvin J. Wachowiak received an M.S. from Winterthur Museum, Art Conservation Program, University of Delaware and a B.S. in studio art from Springfield College. He has been at the Smithsonian since 1989, and involved in projects in the fine arts, anthropology and archeology, and the history of technology collections. He has lectured widely, and his work on protective and decorative coatings resulted in US Patents in 2000 and 2001. Since 2004, he has lead MCI’s use of new and advanced spatial and spectral imaging techniques such as 3D scanning, 3D microscopy, and other computational and digital imaging.
E. Keats Webb graduated in 2007 from the University of North Carolina at Chapel Hill with a BFA in photography. In August 2010, she started her own imaging business focusing on scientific and computational imaging and event photography. She has been an Imaging Specialist Contractor since 2010 working with the Smithsonian Institution including the Museum Conservation Institute, the Freer | Sackler Galleries, the National Museum of American History and the National Museum of Natural History. Her work includes scientific and computational imaging, research and technique development, digital asset management, and video.