… and you’ll be surprised what he or she can do with it. That’s what the National Endowment for the Humanities and the Department of Energy figured. Last year, they staged a competition for “computationally intensive” humanities projects that would draw on the DOE’s High Performance Computing (HPC) resources at Nersc, the National Energy Research Scientific Computing Center at Lawrence Berkeley National Laboratory.
Was the gamble worth it? Yes, to judge by the results on display at the Coalition for Networked Information membership meeting, held in Washington, D.C., this week. Several scholars involved in the HPC competition reported on their supercomputing experiences. Among them were Gregory Crane, editor in chief of the Perseus Digital Library at Tufts University, and David Bamman, a computational linguist with the project, which has been experimenting with computer-enhanced ways to mine a huge digital assembly of classical texts.
David Koller, of the University of Virginia’s Institute for Advanced Technology in the Humanities, gave a nifty talk about creating digitized, 3-D models of culturally significant objects. Mr. Koller is a director, with Bernard Frischer, of the Digital Sculpture Project. As part of that work, Mr. Koller travels to museums and archaeological sites and takes hundreds of photographs of objects—one being the Laocoon sculpture at the Vatican museum—from every angle. Those “raw-scan data” then undergo a complex algorithmic alchemy (that’s the HPC part), converting the pictures into high-resolution images that can be viewed from all angles and deliver an amazing level of detail, down to individual chisel marks. To get the full effect, one needs ScanView, free software created at Stanford University that allows users to virtually examine images that would otherwise be off-limits because of their file sizes or licensing restrictions.
What Mr. Koller does would be impossible without high-peformance computing, he said, but scholars may be able to find what they need close to home. Many universities now have local HPC systems equal to the challenge. According to Mr. Koller, what’s really needed is more support for the human element—people who know how to help computers and humanities researchers talk to each other.
5 Responses to Give a Humanist a Supercomputer …
11333651 - December 16, 2009 at 4:18 pm
Please run spell check on this article and re-post.
mbelvadi - December 18, 2009 at 6:53 am
It’s very unfortunate that taxpayer money has been used to engage in a huge digitization project that requires members of the public to purchase a for-profit product (Microsoft Windows) in order to view the results. ScanView itself may be free, but it only runs on Windows, not Mac or Linux. This might have been reasonable in the 1990s, but with the shifting desktop market, this shouldn’t be acceptable today.
scanviewrox - December 18, 2009 at 7:43 am
Ms. mbelvadi, no public funds were used for any “huge digitization project” referenced in or relevant to the article…what project are you talking about? I’m 100% certain that the developer of ScanView would be more than happy to port the software to Mac & Linux if he had a government grant…
mbelvadi - December 18, 2009 at 11:30 am
“creating digitized, 3-D models” sounds like a digitization project to me. You may dispute the subjective term “huge” but if it needed the HPC, that sounds huge to me. The developers of ScanView can do what they want, but I’m suggesting that the Digital Sculpture Project shouldn’t have tied their government-subsidized (through the use of HPC) project to the ScanView technology which limits access to the results to Windows users.
kintopp - December 18, 2009 at 12:51 pm
At Chicago we use a java version of Scanview to display 3D images of buddhist statues – runs just fine on a Mac and I presume under Linux too. See “Manipulate 3D Scans” at http://xts.uchicago.edu Whether or not HPC per se is really required for the job is a bit up for grabs.. it depends in part on how many people you want to allow access to (how many of) your models at the same time and with what degree of fidelity. It’s certainly not trivial.