In the winter of 2011, I made the four-hour drive down from Chicago to Kampsville to attend the Center for American Archeology Winter Fundraiser and to plan future research with my colleagues. While I was there, CAA research director Jason King let me photograph the model of the Koster dog burial that is on display in the CAA Museum. My goal was to test the then-emerging free resources available for creating three-dimensional renderings of objects from collections of photos and to see how it could help to promote the education and research activities at CAA.

In situ image of Koster dog burial, July 1975 (photo by Del Baston)

In situ image of Koster dog burial, July 1975 (photo by Del Baston)

The model in the CAA Museum is a replica of one of the four dog burials uncovered in Horizon 11 at the Koster Site by members of the Northwestern University Archeological Field Schools and the Foundation for Illinois Archeology (which later grew into the CAA we know today). These dog burials are important for two reasons. First, they were clearly buried intentionally; this particular dog’s grave was furnished with a grinding slab and rubbing stone. Second, having been buried approximately 8500 years ago, they also are among the earliest known domestic dogs in North America and figure prominently in studies that trace the global relationship between humans and domestic animals through time.

As one of many of the significant findings to come out of the Koster Site investigations, I thought that this would be a great thing to add to CAA’s online presence. At the time I was inspired by the emergence of free online resources, specifically Project Photofly -now called 123D Catch- by Audodesk. This was one of several online applications that could take sets of overlapping photographs of any kind and generate a three-dimensional model of the results. This is carried out through a process called Structure from Motion (SfM) an algorithm that measures the spatial distortion in images taken of a subject from different perspectives to first calculate the positions of the cameras and the attributes of the camera and lens. When that is known it is then possible to recreate the surface of the subject.

Camera Positions for Dog Model

Reconstruction of Camera Positions for SfM Model in Agisoft Photoscan

Jason, Taylor Thornton, and I carried the model of the dog burial out onto the back porch of the museum where I could take advantage of winter sunlight diffused by the heavy layer of clouds -good light for photography. We removed the protective plexiglass lid and I took 47 photographs of the model from regularly spaced positions around and above the the dog. I uploaded the model to the Autodesk site, it threw them all together into a virtual 3D model. We received a 123D Catch reconstruction of Koster Dog Burial which we later shared on CAA’s Facebook page.

I had more or less forgotten about the digital model of the Koster dog after we posted the video online, but like many other archeologists, the use of SfM in my work has increased in the last few years. Now SfM is a part of all of my field projects in some way or another, documenting landscapes, excavation progress, monuments, and individual artifacts. SfM and other three-dimensional documentation techniques -laser scanning for example- are now a standard part of the archaeological recording process in the field and the laboratory.

I made the most recent 3D model of the Koster Dog Burial in Agisoft Photoscan on a whim as I was preparing a similar models for an unrelated project (I call this ‘productive procrastination’). Photoscan is a pretty powerful tool and I was interested in seeing what kind of improvement there was over the 123D Catch version. The new model is significantly more precise and of a higher resolution than the earlier model, and the product is quite versatile in terms of the file formats options for export. I was also able to incorporate my measurements of the replica’s frame so that the output is something that can be measured in virtual space.

SfM Modeling of Dog Burial

Two Phases of the SfM Modeling Process in Photoscan. Dense Point Cloud on the Left, Mesh Surface on the Right.

The online popularity of the first Dog Burial model and speed at which it spread is a reflection of the many strengths and advantages of digital representations of field and laboratory observations. I’d like to point out just two of these advantages here. First off, and perhaps most importantly, image-based digital representations are a way to present the public with a rich, sometimes interactive look into the world and work of academic professionals. This is particularly important for archeologists who are often legally obligated to share their results with the public when they work with public money, and ethically obligated to share the information about our shared cultural heritage. Also, as digital data are often easily shared, more people can have access to raw or minimally processed data to carry out their own studies.

The second advantage -and one which I find particularly exciting- is in the potential to improve or revisit digital data (or digitized data) that was collected in the past. Specifically, I’m excited about the power of SfM to turn photographs from any regular camera into precise and measurable 3D models that can be reprocessed and improved upon at a later date as the technology advances. The gap between collection and reprocessing is just four years for this dataset -which is not a long time- but imagine what is possible if we were to mine the photographic record which now extends more than a century into the past!  In fact, many archaeologists have already began to explore the potential for using the historic photographic record to explore past landscapes that have since been compromised by urban expansion and industrial agriculture (corona.cast.uark.edu) and to recreate objects and monuments that have been destroyed in warfare (http://projectmosul.org/).

With our geophysical surveys at various sites throughout the Lower Illinois River Valley and the geomatics track in the ASUFS, digital methods are already a big part of CAA’s  research and education activities. I think you can expect a lot more in the years to come.

Here’s the model again!

Koster Dog Model by Jason Herrmann on Sketchfab

Tagged with →  
Share →