The Traveling CANVAS

Home | About | Calculart at the Dennos | Schedule | Contact

Contents

Abstract
Overview
The Traveling CANVAS
The Art
Conclusion
References

Abstract

    CANVAS, the Collaborative Advanced Navigation Virtual Art Studio is a scalable, reconfigurable display technology for modern art museums. Located at the University of Illinois, USA, the original CANVAS is operating as a 21st century gallery in the Krannert Art Museum. CalculArt, An Exploration of the Intersection of Math and Art was the first curated exhibit in the CANVAS. To broaden the reach of CalculArt in the CANVAS, we committed to a January 2008 gallery opening of CalculArt at the Dennos Museum Center in Traverse City, Michigan, USA. We describe the world's first portable immersive visualization gallery The Traveling CANVAS.
top

Overview

CalculArt occupies a 12 x 20 meter gallery at the Dennos Museum Center, with less than a third of that space dedicated to the CANVAS. Every attempt has been made to find or create examples of mathematical art that exploit similar mathematical concepts using different media. For example, using a particular 3D snapshot of the Romboy Homotopy, a four-dimensional object, one can create a shape resembling an idealized female figure. Three 3D forms of this "Venus" are presented in the gallery, one a PHSCologram using lenticular filters, one a traditional albeit computer generated/printed sculpture and one a virtual, time-morphing version in the CANVAS. Since CalculArt attempts to explore the intersection of math and art, pieces from established artists not necessarily concerned with the mathematical validity of their work and pieces from mathematicians not necessarily concerned with the artistic merit of their work were solicited for the show.

History
The CANVAS can trace it lineage directly back to the CAVE^TM [1], the Cave Autonomous Virtual Environment, developed at the Electronic Visualization Laboratory at the University of Illinois at Chicago, Illinois, USA. Proposed in 1992 and enhanced from that day to the present, the CAVE is an expensive to build and maintain room-sized virtual environment dependant on fragile and expensive liquid crystal shutter glasses to decode the frame sequential active stereo information arriving on the screens at anywhere from 96 to 144 images per second via even more expensive and fragile 3-CRT or 3-DLP projectors. The CAVE is designed for one person to optimally view images that are drawn in real time based on that person's head position in the space, the position being determined at from 100 to 130 instances per second by a matched set of three orthogonally oriented transmitting and receiving coils, a fragile and ferromagnetic-sensitive system prone to malfunction by something as simple as a viewer wearing a hearing aid. Until very recently, a CAVE would require a graphics supercomputer and a team of Unix programmers and system administrators to interpret the ideas of researchers and artists into software code usable by the computer. CAVEs were secured in a room specifically designed for one or specially modified at significant expense to hold one, never to move without great expense of money and labor. The expense and usability challenges constrained the construction of CAVE technology to only the largest research universities, government facilities and a few automotive, aircraft manufacturing and oil exploration corporations.

Challenges
The potential for CAVEs to become a new medium for artists has been recognized from the beginning of the technology. However, several challenges had to be overcome to make even modest inroads in allowing immersive visualization to be even semi-ubiquitous. Replacing the phenomenally expensive hardware with a system approximately 1/20 the cost was described in our first paper "CANVAS: A Virtual Reality Environment for Museums" at EVA Florence, 2006. Our first attempt to minimize the software burden on artists wishing to use the CANVAS was described in "Application Framework for CANVAS: The Virtual Reality Environment for Museums" at EVA London 2006. Further attempts to replace a cluster-supportive, but still somewhat difficult to program system by using a newly created scripting language specifically created for artists was discussed in "Creating Immersive Art without a programmer: The first year for CANVAS, A Virtual Reality Environment for Museums" at EVA Florence 2007. Having simplified the hardware and software burden over the last several years for electronic artists wishing to use immersive environments as a medium for their art, one remaining challenge is the immobility of the structure and requisite supporting systems, making relatively rapid changes in the shape or location of the system very difficult to perform.

Art created for an immersive environment such as the CANVAS can be incredibly compelling, being unbound by the laws of physics constraining sculpture or framed two dimensional works. However, as is the case with much electronic art, pieces are as ephemeral as the electronics used to display them are temperamental. For immersive art to be taken seriously or to at least gain wider acceptance by artists, we have undertaken the project of having that art leave a shadow in the form of sculpture and framed pieces derived from the virtual art in the CANVAS. Additionally, sculpture and framed pieces, long consigned to the physical limitations of walls and pedestals, are being freed from their constraints to appear inside the virtual world courtesy of laser scanners and the software to make the resultant files usable in virtual worlds.
top

The Traveling CANVAS

Hardware Infastructure
Until the Traveling CANVAS, room-sized virtual environments were custom built into the space available. This is true even for the original CANVAS at the Krannert Art Museum, a relatively recent project. A three orthogonal walled structure, typically 3 meters on a side, having rear-projected images, was the standard configuration, so artists had to conform to the constraints of a design created without their input.

However, artists might not want three orthogonal walls, museum directors might not want the space enclosed so tight as to limit the number of attendees able to experience the artwork simultaneously and might object to being asked to remove a structural column to accommodate the projector throw distances, and curators will want to be able to quickly reconfigure the environment during a short show setup time at minimal cost. We therefore redesigned the physical screen structure to make transitions in the overall viewing experience simple to accomplish by making the screen surface, previously one continuous screen to be three discrete screen/frame assemblies with infinitely adjustable sets of angle adjusters at each corner.

Several works presented in the CANVAS were 3D scans of mathematical models from the Altgeld Collection of the University of Illinois at Urbana-Champaign. The original idea was to scan these models using a Shapecam, but that particular piece of hardware was not suited to the task in that the hardware of the scanner was bulky and required a lot of work in terms of human precision to get a proper scan. It needed a lot of space to set up a scan as the lasers needed to focus several feet from the scanned object. Scanning with the Shapecam involved using two lasers to determine and maintain the same focal distance of the camera apparatus as the camera was moved around an object, then taking a pair of pictures at each location with a digital SLR camera mounted to the same structure as the lasers, one picture with a projected precise linear grid overlay for 3D information, one picture for texture. At least eight scans around an object were necessary and the resultant files were electronically stitched together in the Shapecam software. The stitching grid would get lost or completely miss fine detail in an object and rotating the object to capture the parts occluded during the first set of scans was not possible. Texture blending was not a functional feature, leaving obvious discontinuities in image luminance, hue and color saturation due to interactions of natural and flash light around the object. This system however was vastly superior to the previous generation of scanner, the Cyberscanner, which used a single laser to rotate around an object producing a very rough resultant virtual image with no texture and the ability to blind a person not cognizant of the location of the scanning mechanism.

Our choice for obtaining scanned images of sculptural artwork for CalculArt was the NextEngine 3D scanner. This scanner was much more intuitive and turned out to work well, for the most part, to scan the Altgeld models. The physical layout involves the scanner, a relatively small box containing four lasers to obtain depth and position information and a high resolution CMOS camera and flash unit. This system integration is similar to the Shapecam, albeit in a fixed location. A turntable holding the object to be scanned is attached to the laser/camera package and is rotated automatically by the software capturing the images. Connected to a pc with lots of memory, the scanner setup takes about two by three feet of space and can scan objects of moderate size automatically. The software was more intuitive to use, although it had difficulty processing the large amounts of data we fed it. Scanning the models required special care; they were at least eighty years old and made of plaster. Not only were they likely to break, as with any artwork, the oil on one's hands could cause permanent marks on the surface. Because of the fragility of the works, it was difficult to angle the works to capture interior angles, but for objects that could be multiply positioned and not overly incised, well behaved scans and a mastery of the software allowed us to import the resultant files into the CANVAS in a near assembly line fashion.

Printing in 3D has been possible for a number of years with stereolithographic and metal deposition rapid prototyping machines capable of producing an object from an STL file in a single color. Recent advances in rapid prototyping allow creations of objects from numerous file formats. Of most interest to us are formats retaining color information, OBJ and WRL (a VRML format). Whether the file is derived from a 3D scanner or created using software having a scientific or artistic user base such as Mathematica or Maya, realistic model creation demands a printer with high resolution and as broad a color palette as possible, the same as the requirements for 2D printing of artwork rendered electronically. We used a zCorp Model 406 printer to print several of the models in CalculArt. Ironically, this high-tech printer reverts to printing in a media well known to human sculptors, plaster. Pieces took from sixteen to thirty-six hours to print due to their complexity and size, so overnight or over weekend printing sessions were the norm.
top

The Art

The works displayed in CalculArt thrive on the dichotomy that math is not constrained by limits on dimensionality while art must of necessity be presented in a limited number of dimensions.

Scott Carter, Nicholas Duchnowski and Tony Robbin present framed pieces drawn from four-dimensional concepts. While artists still labor with the loss of a single dimension in going from the natural world to a painting and museum attendees are at least familiar with the real world ideas represented, losing two dimensions and attempting to communicate with an audience that cannot ever experience the four dimensions under discussion pose a particularly vexing problem. Consider the challenge of creating art two dimensions down from nature; artists would have only a line of no width on which to draw. The language of trompe l'oeil for four dimensions is only now being formulated.

The vast majority of the pieces presented in CalculArt were three dimensional, either real or virtual 3D, since working in three dimensions affords artists a huge advantage in dimensional real estate for their mathematically based art. Most of the sculpture was produced by fairly non-traditional approaches in plaster, with one piece in bronze by Dickson made through the lost wax method, but from original artwork made on a 3D stereolithographic printer.

Multimodal Art
Two series of artwork will illustrate the diversity of media in CalculArt.

Fractal Crystal #1
Dr. Robert Fathauer posed an interesting question to the CalculArt team: What would you end up with if, starting with a cube, one were to add centered half-sized cubes to each of the available surfaces of the original cube, iteratively to the limits of present computational and 3D printing capabilities? Nicholas Duchnowski took up the challenge. The vitrine-encased object is the end result of printing a >1GB file representing 11 generations of half-sized cubes with the resultant crystal forming a shape and range of colors surprising to even age-hardened geometers. The program that Nicholas wrote to generate the Virtual Reality Markup Language (VRML) file that was sent to the zCorp printer is the same program used in real time for interaction with the various iterations of the crystal by gallery attendees visiting the CANVAS. The virtual object however has the additional property of being able to change the number of iterations being viewed by the simple button press on a gamepad, making virtual art interactive in ways not possible with material-based sculpture.

The Venus Series
In the 1980's, when 3D computer-generated mathematical visualizations were created, they were generally created on graphical supercomputers manufactured by Silicon Graphics, computers often residing at National Science Foundation-funded supercomputer centers across the country. Two such locations shared a common, but geographically-disparate university, the University of Illinois at Urbana-Champaign and Chicago. Professor Donna Cox, George Francis and Ray Idaszak at UIUC and Ellen Sandor, Tom DeFanti and Dan Sandin at UIC collaborated on a PHSCologram titled "Etruscan Venus." Described by the artists as "a video portrait of a Romboy Homotopy, a four dimensional object," the viewer is able to see three of those four dimensions without the aid of stereo glasses due to the partial masking of the multiple images by a lenticular film in front of each image. An additional "Venus" is available in the CANVAS, where the viewer can maneuver the image through various 3D slices of the 4D object. Stewart Dickson plays off the Venus theme with a sculptural piece Botty Shelly, also in the CalculArt gallery. top

References

[1] C. Cruz-Neira, Sandin, D., DeFanti, T., Kenyon, R., Hart, J., "The CAVE: Audio Visual Experience Automatic Virtual Environment", Communications of the ACM, vol. 35, no. 6 06/01/1992, pp. 65-72
top

The Traveling CANVAS is developed by the ISL at the Beckman Institute at the University of Illinois at Urbana Champaign.