 |
|
For a long time the TeleType was still. Then it started clacking. It wasn't printing a news bulletin, though; it was printing lines of spaces and an occasional character, line after line of apparent random noise. This continued for a while when I realized that the machine was building an image, a portrait of JFK with a caption at the bottom taken from his inaugural address, "Ask not what your country can do for you, ask what you can do for your country." Someone down the line had punched this picture into paper tape and transmitted it on the wire service. I tore off the page and took the picture home to be framed and displayed, a vestige of the era. I still have it somewhere, but after 34 years and nearly a dozen moves, JFK's portrait is lost, stored in an unopened carton somewhere among the stuff that you never get around to unpacking. I hope I find it someday.
That picture has always fascinated me. It uses very few characters from a small set of uppercase letters, digits, and punctuation to render its image, yet the likeness is unmistakeable. I wondered how someone could have conceived the correct combination and placement of characters to suggest so vividly the face of a person and then construct the portrait by using a medium never intended to render an image. I wondered if they might have used a computer, a curious possibility for the early '60s.
In the 1970s I walked into an office and saw what appeared to be Abraham Lincoln's portrait hanging on the wall. You can see that picture at http://www.ioc.net/~artboy/pages/where/harmon.html. On closer inspection, I found that what seemed to be a portrait was really something else, in this case a matrix of large squares in black, white, and shades of gray. Who did that, I wondered, and how did they do it? Later I learned that the picture had been published in the November 1973 issue of Scientific American along with an article by Leon Harmon entitled "The Recognition of Faces." You can find the same image in a Salvador Dali painting at the museum dedicated to his work in St. Petersburg, Florida. You can view that picture at http://home4.swipnet.se/~w-40823/Dali.html.
The painting follows the tiled architecture of the original and uses the original as one of the tiles. Dali's paintings are the ultimate extension of the art of rendering suggestions of images from other visual components. They defy description; even high-quality prints in books cannot capture the subtleties of tone and color that Dali used to build images within images; you have to see them in person. Some of them are several stories high. In his book Dali (Thunder Bay Press, 1994), Paul Moorehouse quotes Dali's reference to such visual illusions as being "consubstantial with the human phenomenon of sight," which Moorehouse relates to "the mechanics of perception: the ways in which the human brain decodes and interprets visual information." You can view and download bitmapped images of some of Dali's work from http://www.nol.net/~nil/dali/gallery5.html.
The original picture of Lincoln is a mosaic that could have been generated by a computer, but was probably done with a photographic process. You can do something similar with any photo. Scan it into a paint program that has good image resizing and screen area capture functions. (The shareware Paint Shop Pro is such a program.) Reduce the image to icon size. Capture and expand the icon. Computer-generated images are composed of picture elements--pixels- -which are different colored dots organized to suggest an image. The squares that suggest Abe's face are pixels of extremely low resolution, which means that each pixel is very large. The farther away you get, the more effective is the illusion that you are looking at a photographic portrait rather than just an assortment of tiles.
The art of computer-generated mosaics of pictures made from tiled pictures is fairly new. Examples of the art are available in a book titled Photomosaics, a compilation of images created by computer programs written by Robert Silvers. Michael Hawley edited the book and wrote the accompanying text, which includes appropriate quotes from various people. Hawley is a professor at MIT. Silvers was one of his students and is now president of a company that makes logos and illustrations for individuals, corporations, and publications. The authors combined the NASA term "photo mosaic" to coin "photomosaic" to apply to the process and the result. The book consists of over two dozen such images. Many of the images are portraits, including renderings of the faces of Jesus Christ, George Washington, van Gogh, Abraham Lincoln, Marilyn Monroe, Elvis Presley, Bill Gates, Madonna, Al Gore, and Yoda. Each image is a mosaic of related images. Lincoln's portrait consists of Civil War photos; Elvis is composed of postage stamps; Madonna is a montage of material things; Bill Gates is made of money.
The photomosaic software begins with a scanned image of what is to be recreated in the mosaic format. Which means that the final pictures are not really computer-created art; they are computer-replicated art wherein the replication is an approximation of the original, the interesting part being that the approximation uses hundreds of subject-related images. The software divides the original image into tiled regions and assigns properties to each region with respect to color, brilliance, shape, and so on. Then the software searches its database of contributing images, images that are candidates to become tiles in the final product. The characteristics of each database image are compared to those of each tiled region in the original picture. This process is processor intensive; the images are not of equal dimensions, they overlap, and the matching algorithm has to pan the candidate tile images around looking for a good match. The software needs high-power hardware and lots of storage for the typically 2000 images that the programs analyze to compose a single mosaic. Silvers wrote the program in C for the Silicon Graphics Onyx. He chose C because he says he is "not a big fan of C++." (He's young, though, and there's still time for him to see the light.)
Creating a mosaic is an iterative process involving the computer's output and manual refinement and tuning of the image by a person. The hand and eye of the artist are not made obsolete by the software; human interaction is a necessary part of the process.
The software that produces the images is not publicly available. Silvers wrote the programs to use as tools in his business. He says that he does not wish to operate a software company. Can't blame him there. Between the heavy iron requirements, the immense image database, and the need for an artist's influence on the output, imagine the tech support problems. Silvers is registering "photomosaic" as a trademark and patenting the photomosaic algorithms.
Not all of Silvers' work is publicly available for viewing. One of his early commissions while he was at MIT was for Penn Jillette of the comedy/magic team of Penn and Teller. The picture, a mosaic of pornographic images downloaded from the Internet, can be viewed only in Jillette's Las Vegas home. Jillette wrote an endorsement of the process that was to be a foreword to the book. The publisher decided not to use it possibly because of its irreverent tone and its subject material. I think it is funny. You can read the foreword and see view some of the images at the Runaway Technology Inc. web site (http://www.photomosaic.com/).
Photomosaics is a delightful coffee table book that comes complete with a credit-card sized pocket magnifier so you can look closely at the tiled images. I highly recommend this book, particularly to anyone interested in computer art.
-- Al Stevens (astevens@ddj.com)