This page provides an overview of the workflow we use at the Gabii Project excavations, starting with taking the photos of a stratigraphic context, through creating a 3D model, linking to other data collected at the site, and designing the tools we will use to interact with the data and models. If you're not familiar with the use of image-based modeling in archaeology and the Unity3D game engine, we recommend reading this page before diving into the 3D and ARK sandboxes.
Our digital data are collected as part of the excavation process. When we uncover a new stratigraphic context, we survey its limits using a total station. For complex stratigraphic contexts, we collect a sequence of photos and survey control points which are used to generate a geo-referenced 3D model. The collection of this basic data is integrated into our normal documentation strategy.
We use Agisoft PhotoScan, a software package designed to carry out image-based modeling, to create models of each individual stratigraphic context. The software searches for points which correspond between images in a collection (essentially a modified SIFT algorithm). Once corresponding points are identified, the software searches, starting from these points, for further matches. Accepted matches are used to construct the mesh surface of the 3D model. The color information from each pixel is used to assign color or texture-color values to each vertex or face of the mesh. Once the mesh has been constructed, the targets in the model are assigned coordinates which were surveyed in the field, fixing the model's location. Georeferenced models are cleaned using mesh processing software including Meshlab and Blender. You can learn more about a (now slightly dated) workflow for using Agisoft Photoscan in archaeological projects at CAST's GMV website.
The development of the 3D modeling content at Gabii has run in parallel with the development of a web-based database which houses and makes accessible the rest of the project's digital record. The Gabii project uses a customized version of the ARK (Archaeological Recording Kit - developed by LP Archaeology) to manage descriptive data, and entries in the Gabii Ark can be accessed directly from Unity3D scenes. You can access a basic version of Gabii's newARK, both directly and via links in the Unity scene.
Individual models are assembled into scenes designed to help resolve a question, or to understand a stratigraphic situation. In addition to including the stratigraphic contexts themselves, we can experiment with different reconstructions in these scenes by adding elements not recovered in the excavation, but which we think may have been present during the construction, use or abandonment of a structure or area. We assemble these scenes using Unity3D, a software program primarily intended for the design of games, and now widely used for 'serious games' in a variety of academic contexts.
Just as there is an important difference between publishing data and publishing a narrative, argument or synthesis, there is an important difference between publishing a digital model and publishing a narrative, argument or synthesis through that model. The way in which we can interact with the scenes - the means provided to interrogate the models and explore the information presented through them - is as important as the models themselves. Interactivity includes everything from highlighting a context when clicked to the ability to query for contexts with a particular property, or to link from narrative text to 3D scene to database entry, or to guide a reader through different parts of the model in a specific sequence. This kind of interactivity is essential for creating models and scenes which will integrate well with other modes of academic communication.
The Harris Matrix is widely used in single-context excavations to visualize relationships between stratigraphic contexts. Establishing these relationships is essential to the creation of relative chronologies and the delineation of activities and phases. Network analysis tools like Gephi and Cytoscape provide us with flexible, web-friendly means of generating network visualizations to give researchers both familiar and new views of the networks of stratigraphic relationships, which are the fundamental building blocks of excavation based narratives.