One of the piece of research I’ve been working on finally came to fruition in a busy series of experiments and write up in the last two weeks. The short study (on which I presented a poster at Digital Heritage, York) aims to assess user interaction with three interfaces for displaying landscape data. The purpose is twofold; to improve pedagogical approaches to archaeological landscape analysis using lidar-derived terrain models and to provide feedback to inform the design of a public interface highlighting the historic landscapes of areas such as those VMS551 and I surveyed back in March that are challenging to access and interpret.
Although we are increasingly dealing with terrain models and imagery deriving from airborne lidar survey, in the archaeological community we almost always view these data in 2D via a Geographical Information System (GIS). There are distinct advantages to this approach, including the ability to accurately map potential features and easily orientate the data with respect to other resources such as historic and modern mapping and aerial photography. However a 2D interface doesn’t give much of a sense of the general topography of the landscape or micro topography of the features of interest within it. Having taught a number of classes on the topic, I suspect that this 2D approach to 3D data hinders the speed with which the uninitiated can get to grips with and use terrain data in their historic landscape interpretations. So I set out to discover what difference exploration in 3D virtual environments made to their learning and discovery process.
Through the summer I developed a workflow to process lidar data and derivatives from the GIS files where they so happily lived throughout my PhD to 3D models that could form the basis of a virtual environment. This was a little bit convoluted at times but I was able to derive a process from GRASS to Paraview to Meshlab that resulted in a workable .obj file with multiple textures for display.
Fortunately here at Duke we have the pièce de résistance of virtual environments – a 6-sided cave projection system known as the DIVE (Duke Immersive Virtual Environment), and I was lucky to have the support of the director of facility, Regis Kopper, to develop my experiment. The DIVE provides a fully immersive 1:1 scale, landscape exploration experience, which is very impressive in terms of the sense of place, perspective and scale you feel walking through the landscape models. Even for a hardened researcher like me, it was a thrill to virtually stand atop a Bronze Age barrow I had climbed in real life and survey the landscape. A huge shout out has to go to the ever-generous David Zielinski for making the interactive aspect of the virtual worlds happen – I would certainly have never got up to speed on the software fast enough to go it alone!
Of course not everyone has access to such a facility and by far the most distributable 3D environment is some form of web browser (either native 3D or via a plug-in), so in this experiment we also developed a web-based environment. Sadly due to time constraints, this was not built in an open-source platform, but in Virtools (which currently powers the DIVE applications). Given time I hope to develop an X3D / OpenSceneGraph version instead, but below is a short introductory video which will give you a feel for the environment.
I was able to persuade 12 volunteers, some with experience of landscape archaeology and some without, to come and participate in the experiment to compare all three environments directly. After a introduction establishing the experimental method and explaining the data, each participant was set the task of looking for topographic features that might be archaeological within three different landscapes and to narrate their explorations and interpretations. The results will be compiled and presented in a research paper, but will also feedback to inform the development of a fully-fledged interactive environment for upland historic landscape exploration that is one of my longer term research goals.