Image courtesy of A. Zahner Co. 

The concept of covering the largest copper-clad building in the United States with millions of bumps and holes came from the architects Herzog and de Meuron.  The mechanical process and sheet-metal skills came from A. Zahner Co.  The software that made it possible to manage massive amounts of data and complete the job in this lifetime came from the keyboard of Eric Wilhelm.

The abstract texture of a forest canopy was created from high-contrast digital images.  These were then mapped onto the building surfaces and "pixelated" into a finite set of depths or diameters.  This software allowed the architects to airbrush and smudge the texture with an image editor, and then submit the images for automated batch processing. 
Image courtesy of A. Zahner Co.  The hands-off process allowed the architects to iterate the extraction process as many times as needed, giving them complete artistic freedom over what ultimately turned into g-code for driving the numerically controlled hydraulic punch that produced each of over 8000 unique panels.

In addition to hundreds of other concerns, the bumps also had to be placed in an in-out checkerboard and the consistency of this alternation checked around the entire building and across image seams.  Unfortunately, because of the complexity of the folded and tucked surfaces, there is one corner where the traversal of the building comes out odd instead of even.  See if you can find it when the museum opens in October 2005.  To achieve the checkerboard with a uni-directional punch, two data files per panel were required to drive the numerical equipment.  These had to be accurately registered to each other so that the 1/32" tolerances could be maintained when flipping a 39" wide copper sheet.

While this was an absolutely one-of-a-kind project, the software was constructed in such a way that many of the components were able to be reused.  One of these is my CAD::Drawing library, which A. Zahner Co.  allowed me to develop as an open-source application as reciprocation for the value that they were getting from Linux, Perl, Apache, and other open-source software.

I also drew on the experiences from this project to develop a suite of programs which handles automated engineering of sheet-metal cladding and flat-pattern processing.  These programs were built using Unix-style concepts to provide flexibility, expressiveness, and power.  To date, this system has been used on some of the final parts of de Young and at least three other completely different cladding systems.

This is the sort of thing that births buzzwords.  It's one thing to talk about reevaluating paradigms and leveraging synergy to efficiently automate your core processes and reduce errors while achieving better return on investment.  To actually "do it" requires careful insight, skill, hard work, and a little bit of magic.

OSCON 2005

I gave a lightning-talk presentation to an audience of perl developers at OSCON in August 2005.  Click here to read it.

Pictures

I'll have to get some pictures of my own when I get a chance to go down to SF and see the museum.  For now, I've got some that were provided by A. Zahner Co.