Harrison Eiteljorg, II
One of the CAD matters that has concerned me for some time is the proprietary nature of the file formats we are obliged to use for our CAD models. For example, AutoCAD®'s DWG format has become a de facto industry standard, but it is a proprietary format that Autodesk controls absolutely. In fact, it was clear that Autodesk changed the DWG format ever so slightly when IntelliCAD® was released a few years ago so that there would be minor problems for users of IntelliCAD who tried to create DWG files. (See "Software Comment: IntelliCAD," H. Eiteljorg, II, vol. XI, no. 2, Fall, 1998; http://csanet.org/newsletter/fall98/nlf9807.html.)
Changing the specifics of a file format is an easy way to cause trouble, but that is not the issue that most worries me. I am, in fact, more worried about the fact that a given file format may represent information in ways for which there are no analogs in another file format. That is, the simple geometry and layer divisions in any CAD file are the only critical things that must exist in all file formats and therefore must be translated when a file is migrated from one format to another. However, other information stored in a given file format -- for instance, database linkages or rendering materials and lighting choices, made in the process of creating renderings -- are buried somewhere in the file, though a user need not know where. Such information may or may not be used in other programs; so a translation may or may not include it.1
Not only is it difficult to translate non-geometric information from one file format to another, that non-geometric information is not the crucial part of the model. That is, the typical scholarly model is created as a way to present the geometry, with layer distinctions to be sure, but not a vehicle for rendering exercises or database information.
Some CAD models, of course, are made as much for the visual results as for the geometry -- Kent Larson's models of Louis Kahn's unbuilt structures come to mind -- and others include substantial information in the form of attached database files. Putting those aside for the moment, though, I now believe it to be important that standard models intended for scholarly use be prepared carefully so as to be devoid of information other than geometric objects, layering, and text that may be part of the model, not in external files. Colors, textures, lighting information, proprietary attachments to external data sources, and any other information not visible in the model should be omitted. I am suggesting this, because I believe it will result in more long-lived -- and therefore more useful -- CAD models. Without extraneous materials, CAD models will be far easier to migrate into new formats as those formats evolve; as a result, the models will be retained more easily and will remain accessible longer.
What about rendering information? When rendering information is important to the model-maker, it should be retained in one of two ways. One, two separate CAD files can be prepared and archived, one with the geometry and layering information alone, the other with the added information required for rendering. So long as migration2 routines can translate the rendering information, both files will continue to be useful. However, should the rendering information become an impediment to continued data migration, the basic geometry would not be lost along with the rendering information. Two, separate instructions for rendering could be saved along with the CAD file containing the geometry and layers. Those instructions might be in simple text files describing the process or in macro commands suitable for use in the CAD program used for the modeling. (Of course, in that case the macro might have to be migrated as well.) Either choice should make it possible to produce renderings of equivalent quality, though both may make it more difficult for an individual to produce exactly the same renderings that were originally produced by the model maker.
Attached database or other textual information can be dealt with in a similar fashion -- by separating it entirely from the CAD model. A companion article in this issue of the Newsletter ("Linking Text and Data to CAD Models," by Harrison Eiteljorg, II) explains that process in some detail and provides additional reasons for separating the model from related data files.
Operating on the basis of these proposals (as the Archaeological Data Archive will), archiving CAD models should become much safer and more reliable. Checking for proper translation of geometric information and layering is relatively simple; so the process would not only be safer and more reliable, but checking for errors would also be better. Additional information in the form of rendering information or attached databases will still be part of the archived package, but they will not complicate the CAD model itself. In the future, if CAD and GIS software merges, as so often predicted, it may be desirable to modify this archival procedure, but I believe that the current state of the technology argues strongly for this safe and simple approach to CAD model-making and archiving. This proposal will also make the choice of software for model-making less important. If only layers and geometry are to be preserved, the issue of file format becomes much less important, and that is certainly one of the matters that has complicated software choices in the past.
-- Harrison Eiteljorg, II
To send comments or questions to the author, please see our email contacts page.
1. As noted in earlier articles, "Review of AutoSurf® - Plus Some Ruminations about File Formats for Scholars," Harrison Eiteljorg, II, vol. IX, no. 3, November, 1996; http://www.csanet.org/newsletter/nov96/nl119604.html and "A Problem with AutoCAD Files," Harrison Eiteljorg, II, vol. IX, no. 4, February, 1997; http://www.csanet.org/newsletter/feb97/nl029712.html, there can also be problems with the translation of geometric information from file to file; so the absence of a single, standard, non-proprietary file format for CAD creates many problems. Return to body of text.
2. This process of preparing two files is not so simple as it may appear. At least in my experience, it is not possible to strip out the rendering information once it has been introduced. Therefore, the basic file with geometry alone must be carefully maintained, and all rendering work must be done in another file. Return to body of text.
For other Newsletter articles concerning the applications of CAD modeling in archaeology and architectural history, the ADAP and issues surrounding digital archiving, or the use of electronic media in the humanities, consult the Subject index.
Next Article: AutoCAD® Drawings as Computer Files
Table of Contents for the Winter, 2002 issue of the CSA Newsletter (Vol. XIV, no. 3)
Table of Contents for all CSA Newsletter issues on the Web
|CSA Home Page|