IMG MGMT: 20 Archetypes

Guest post by: KEVIN ZUCKER

The Stanford Bunny. Three-dimensionally scanned from this terra cotta garden decoration in 1994, the Bunny model has since become one of the most commonly used tests for a wide variety of 3D research projects

[Editors note: IMG MGMT is an artist essay series highlighting the diversity of curatorial processes within the art making practice. Kevin Zucker, today's invited artist is represented by Greenberg Van Doren in New York. He has had numerous solo exhibitions internationally, most recently Standards and Anomalies at Arario Beijing, where he showed work that incorporated imagery discussed in this post. His next project will be a two-person show with Hilary Berseth at Eleven Rivington (NY) in October.]

“These are the archetypes of a new world. They don't need you or me. They're representation for machines.”¹

I've been collecting documentation of objects and images used in the development and testing of digital imaging and 3D modeling technologies for a few years, and have thousands of them stashed on a hard drive and a list of links to databases with literally hundreds of thousands more. The twenty presented here are representative highlights that I think epitomize the inscrutability, banality, anachronism, and the straightforwardly artless presentation that characterize most of the collection. Those qualities, contrasted with the weird aura possessed by these analog “originals” of digital representation, make for the unsteady balance of gravity and absurdity that first got me interested in collecting them.

While the “end user” rarely sees any of these images or objects, a handful of them (Lena, the mandrill, the Utah Teapot (all below), the Stanford Bunny (above)² and the Cornell Box (below)³ ) are well known to the point of being iconic within the digital imaging research community. They have even become the subjects of inside jokes between programmers and animators: 3D models of the Utah Teapot are hidden in Pixar’s Toy Story, a screensaver that comes as part of Microsoft Windows, and the Simpsons episode where Homer stumbles into a computer-generated “Third Dimension”⁴

The Utah Teapot. Modeled in 1975 from this original, the Teapot has become another standard test object used in a very wide variety of 3D applications

Lena. Scanned in 1973, The Lena image has since become the most widely used standard test for a variety of image processing applications

The Cornell Box. A physical model photographed under controlled lighting conditions, the box is intended to be used as a standard for comparison when judging the accuracy of lighting in 3D renderings

The Mandrill. Another very widely used test used for image analysis and computer vision research

Lena, probably the best known of all of these images because of its ubiquity as a standard quality test, is the face of Playboy “Playmate of the Month” Lena Sjooblom, originally cropped from the centerfold in 1973 by engineers at USC⁵. It is still in wide use today in spite of technological advances, past controversies over copyright, and ongoing questions about the politics of its source and content⁶.

The centerfold source of the Lena image.

Lena in use in compression tests

The rest of the images and objects shown here are more obscure, taken from data sets that are freely distributed for research purposes. Some of their applications include: image quality tests; benchmarks for the measurement of processing performance; hardware, software and printer self-tests; the development of synthetic lighting techniques for rendering 3D graphics and animation; statistical shape analysis; the modeling of traffic patterns, and samples for training facial recognition software and robotic/machine vision.

[From the top left] The Mandrill image, Utah Teapot and Cornell Box models in use, and The Stanford Bunny model

The Bunny as used in an animation

These fields of research have developed rapidly and become important influences on our culture. Virtually everything produced for the mass market is now designed in 3D software that made use of a Bunny or Teapot test model at some point in its conception. From the compression of the jpegs we see online to the CAD software used to design the monitors we view them on, the results of research using these test images and objects have been seamlessly integrated into many aspects of daily life.

An image from a standard stereoscopic vision test set

A Standard test image

If you have any inclination toward science fiction, it’s appealing to view the realm of 3D modeling and digital representation as a discrete and clearly defined universe, a parallel reality to our own. In that universe, the objects and images shown here would constitute some of the Platonic forms upon which everything else was based. Even if that sounds overstated, it’s safe to say that the archetypal images and objects that come from our analog world (except for the “as used” images above, the photographs shown here are straightforward documentation of things from the physical world) structure a digital domain that in turn exerts an influence on the way we see and make things.

Another image from a standard stereoscopic vision test set

From the IMM face database. Images from this database are used for testing and development of software for applications like facial recognition and character modeling

While the researchers’ original choice of some of these images as standards may seem arbitrary or eccentric, it’s important to remember that they were never intended to function as representation in the normal sense of the word; they aren’t meant to communicate something about the world we live in. In most cases, their selection was a function of convenience paired with the image’s appropriateness for testing very specific software or hardware capabilities. For example, Greg Turk, who 3D-scanned the Stanford Bunny in 1994, bought it from a local Palo Alto home and garden supply store because the terra cotta material was “red and diffuse” and its geometry was not particularly complex⁷. The Lena photo was scanned because the engineers had been looking for an image of a face that was “glossy to ensure good output dynamic range,” and “just then, somebody happened to walk in with a recent issue of Playboy.”⁸

Standard test image.

Image from a data set of meat photographs used in statistical shape analysis.

Image from ALOI database, a collection of 1,000 small objects recorded with systematically varied viewing angles, illumination angles, and illumination colors for each object

Image from a robotic vision training sequence animation (dead link), a frame from (I think) a setup used in training robots to recognize and remember visual background material

Image from hand sequence data set used in statistical shape analysis

Standard test image used for image analysis and computer vision testing

Image from a standard stereoscopic vision test set

These objects and images (and others like them) have a profound but hidden impact as the building blocks of an increasingly complex vocabulary of digital forms. As future developments in imaging technologies expand on their current capacities, these sorts of mundane archetypes will stay, at least metaphorically, ingrained at their basic levels. In the same way that the 19th century pop song “Daisy Bell” remained at the core of the “dying” HAL‘s regressing memory in 2001, the visual artifacts shown here will remain part of the evolutionary memory embedded in the methods machines use to process and describe visual information.

Image from a traffic sequence used in computer vision testing and modeling traffic behavior.

Synthetic lighting demonstration. 1992 demonstration of synthetic lighting technique used to accurately modify the lighting of a scene after it has been photographed. Lighting controls in the rendering of 3D models work according to these principles.

Image from machine vision test sequence. From a frame in a sequence used for image processing, image analysis, and machine vision research

Another Image from the ALOI 1,000 small objects database

1. Hilary Berseth, email [↩]
2. Turk, Greg, “The Stanford Bunny”, August 2000, Georgia Tech College of Computing, Atlanta, GA, <http://www-static.cc.gatech.edu/~turk/bunny/bunny.html> [↩]
3. __________, “History of the Cornell Box”, , January 1998, Cornell University, Ithaca, NY, [↩]
4. Lindberg, Kelley J.P., “Pioneers on the Digital Frontiers”, Continuum: The Magazine of the University of Utah, Winter 06-07, The University of Utah, Salt Lake City, UT, [↩]
5. Hutchinson, Jamie, “Culture, Communication, and an Information Age Madonna”, IEEE Professional Communication Society Newsletter, vol. 43, no. 3, May/June 2001, pp. 1, 5-7, Professional Communication Society of the Institute of Electrical and Electronics Engineers, NY, NY [↩]
6. ibid. [↩]
7. Turk, Greg, “The Stanford Bunny”, August 2000, Georgia Tech College of Computing, Atlanta, GA, < http://www-static.cc.gatech.edu/~turk/bunny/bunny.html> [↩]
8. Hutchinson, Jamie, “Culture, Communication, and an Information Age Madonna”, IEEE Professional Communication Society Newsletter, vol. 43, no. 3, May/June 2001, pp. 1, 5-7, Professional Communication Society of the Institute of Electrical and Electronics Engineers, NY, NY [↩]