As we continue scanning the Turner Collection, we are also working toward establishing a method of Landmark (LM)/semi-Landmark (sLM) application that is replicable. We have, with the help of Creaform and Geomagic, made some substantive progress in this direction, and are now beginning to test our results.
To date, we have been using a host of digital tools and programs to apply landmarks and semi-landmarks to ceramics; however, one of the more challenging aspects has been that of replicability. While in our minds there is only one landmark that is consistent throughout the entirety of the ceramic assemblage, the central base (CB), it is the semi-landmarks that have been of the greatest interest. Since we want to be as consistent as possible with landmark placement in order to ensure that the sLM’s are placed in comparable spatial locations, we decided it would be best to begin sLM placement at the widest profile on the ceramic vessel.
This invariably led us to another question – how to define the widest point of the vessel? After several discussions with professionals in the fields of engineering and reverse-engineering (and the folks at Geomagic), we were able to find a method of defining the widest point on each vessel.
We start by adding a vector along the revolving axis of each piece, which takes into account the entirety of our dataset (in our case, a mesh). This vector is used to place the single LM at the central base by using a projection of the vector and the poly-vertices of the mesh. The result is a LM that is replicable in two short steps. Once this LM is placed, we begin to create reference geometry on the vessel surface that will be populated with sLM’s.
We then identify region groups (cones, cylinders, revolutions, planes, spheres, etc.). These regions are used to insert a plane on the base of the vessel. The base was chosen because it is consistently the most robust part of the vessel, and is likely the piece that is best preserved (to add to this, the rim of Caddo vessels is often sloped – this is another topic of inquiry that we are currently exploring). It is possible to use the rim, but with the fragile nature of that part of the vessel, it has often sustained some manner of damage. By using the basal plane, this allows us to orient the vessel along the same plane that it would normally sit when atop a table. Further, we assume that this position of the vessel is more consistent with the manufacturing ideals of the original maker(s).
The plane is then used as the basis for a mesh sketch, where it is employed to create a digital silhouette. The CB is the central point of the mesh sketch, which will subsequently be used to extrude a circular surface around the vessel. From that surface, Design X calculates the deviations, and identifies (consistently) the same point of the vessel that is farthest from the CB (not forgetting that the silhouette is based upon the orientation of the basal plane).
Once identified, we apply a second plane (pick point and coplanar axis), where the coplanar axis is the vector, and the pick point is the point that Design X identified as being located farthest from the CB. This widest part of the vessel is the first to be populated with sLM’s once the remaining reference geometry is added.
Results of five separate applications of reference geometry used to identify the widest profile of the vessel.
The application of the remaining reference geometry is more subjective for our consensus configuration, and we are currently running a series of tests to identify how many profiles to use, how many sLM points to add to each profile, and whether there is a plateau (as suggested by a recent biological study) in the number of LMs/sLMs used where statistically-significant results might be achieved (we’re searching for the point of diminishing returns).