Vol. XII, No. 3
CSA Newsletter Logo
   
Winter, 2000

CAD Data Segmentation - A Crucial Tool

Harrison Eiteljorg, II


More and more scholars are making use of CAD for their projects. The growth in use has been slow, but there has been real growth. As I have watched the growth, however, I have realized that new CAD users have been of two distinct kinds. Some have begun using CAD to satisfy needs for good illustrations. That is, excavators and other scholars want to be able to show what a site or building looked like at a particular time; so they turn to CAD models - often in conjunction with other tools that are better suited to the production of high-quality images.

Other new users have taken up CAD as a documentation tool, seeing it as a better way to document three-dimensional reality and as an aid to analysis. Thus, some are primarily interested in CAD as an analytic tool, while others see it first as an illustrative one used to generate renderings and virtual worlds. Neither use excludes the other, but there are different emphases that can incline one toward or away from building a model that is useful for both purposes. That is a matter of concern; no model should be a single-purpose model. A good CAD model should be useful as a base for illustrations and as a base for analysis; there is no inherent conflict.

Unfortunately, those using CAD for documentation have often missed one of the most important features of CAD for documentation and analysis, and those using CAD for illustration are even less likely to use this important feature: dividing models into segments for analytic purposes. The segmentation of the model provides one of the most important analytic tools available to CAD users; it deserves emphasis here.

As many Newsletter readers know already, CAD models may be divided into any number of segments. Those segments can be conceptually, spatially, and/or chronologically distinct from one another, and dividing the model carefully into appropriate segments can provide significant analytic benefits. For instance, different phases of a site or structure can be seen individually or together; in situ material can be seen with or without reconstructed portions of the site or structure, and even different scholars' reconstructions can exist within one model while remaining separate from one another.

Fig. 1 - Segmented drawing of the Older Propylon

This can be a difficult concept to put into words; so the drawing in Fig. 1 is an attempt to express the idea. The drawing shows a portion of the material in the upper courtyard of the older propylon on the Acropolis in Athens. The drawing, as assembled in the center, consists of many individual segments or parts, each of which is shown separately around the perimeter - stones from different periods in the history of the upper courtyard, reconstructed portions, stones of different material (marble and the soft local limestone called poros), stones used for different purposes (steps, wall veneer, cross wall, etc.), stones in situ versus in situ but in secondary use, versus stones no longer in situ at all. Even the cracks on the wall revetment blocks are in a separate data segment (since they were, presumably, not there originally). Note that the groups of stones are distinguished from one another by more than one of these defining characteristics.

Segmenting a CAD model in this way brings crucial functionality to a model for scholars. Once the model has been completed, the user can call up segments (or suppress them) according to any of the criteria - or, more usefully, multiple criteria. For instance, a request for all in situ marble blocks of the last period would have brought up all the portions of the drawing except the holes, cracks, hypothetical reconstructions, and the poros. A similar request for the in situ marble blocks of the final phase ONLY would have omitted all but the anta blocks; all the other marble was used in at least one prior phase. The most valuable uses involve multiple criteria. In this example, for instance, it is telling to call up the material from the final phase that is marble, omitting the limestone and, at the same time, to highlight the material in secondary use position, as shown in Fig. 2. Such a view suggests what, in fact, is the case - that the highlighted blocks were originally positioned with the rest of the marble blocks as shown in Fig. 3.

Fig. 2 - Older propylon - with
poros removed and secondary use
marble anta blocks highlighted.
Fig. 3 - Penultimate phase of
older propylon - with anta
blocks moved from current
position to prior position.

The point here, of course, is not that the archaeology of this example can be reconstructed in a certain way. Rather, the point is that segmenting the model effectively helps in the analysis of the structure, helps point the way to the conclusion.

The fact that a model has been segmented carefully and with helpful analytic categories does not necessarily mean that it will be possible to access the data segments easily and accurately. That is, segmenting a CAD model is one thing, designing the segmentation system so that a user can easily elect to see only certain parts of the model is another. In the example we have been using, for instance, AutoCAD® drawing layers have been used as data segments, and they have been named according to the CSA Layer Naming Convention (http://csa.brynmawr.edu/web1/csalnc.html). (The current URL for the layer naming conventions is www.casnet.org/inftech/csalnc.html - 11 July 2000.) Using that convention (with programs that permit layers to be turned on and off using common computer search procedures) maximizes the benefits of segmenting CAD models. (A simplified version of the layer naming scheme is also used in CSA's AutoCAD tutorial on the Web: http://csa.brynmawr.edu/web1/tutorial.html. Its current URL is www.csanet.org/inftech/tutorial.html - 11 July 2000.)

The CSA Layer Naming Convention has been on the Web for some time; it was developed and refined over a period of years. The first discussion of the need for a naming system, in fact, was in the very first issue of the CSA Newsletter, and suggestions and revisions appeared regularly until 1993, when the current system was proposed. Nonetheless, CAD models are still frequently created without careful consideration of the data segmentation. In addition, there are CAD programs that cannot manipulate data segments in this way. I am writing about this issue again now because it seems a good idea to remind those who are relatively new to the use of CAD that there is an effective system for naming and accessing data segments by analytic criteria. The system has been developed and refined in real-world use; it can easily be adapted for use on new models. At the same time, there may be other systems that are equally helpful, and we at CSA are eager to know of any system that will help scholars use CAD data segments efficiently and effectively. Articles describing such systems are very welcome.

This is also an important issue because those who come to use CAD for archaeological work without much training in CAD often do not concern themselves with segmenting the model at all, much less doing so according to analytic schemes. They may not see the potential, particularly if they have been shown how to segment the model but not how to use the segments in an analytic fashion - as is most likely the case when learning CAD from the usual training sources. However, models that do not include effective layering schemes for analysis rob potential users of important information. Anyone using CAD for scholarly work should be taking advantage of this potential.

-- Harrison Eiteljorg, II

To send comments or questions to the author, please see our email contacts page.


For other Newsletter articles concerning the applications of CAD modeling in archaeology or uses of electronic media in the humanities consult the Subject index.

Next Article: Web Design: What is the Purpose of a Site?

Table of Contents for the Winter, 2000 issue of the CSA Newsletter (Vol. XII, no. 3)

Master Index Table of Contents for all CSA Newsletter issues on the Web

Return to CSA Home Page