The Social Art Internet Art Project

Back in 1999, I began to conceptualize an idea for an Internet art gallery where upcoming artists and musicians could display their work. I built a proto-type, enlisted a handful of artists, and in November of 1999 launched SocialArt.com. Social Art was, at best, a great portfolio. Without mechanisms in place to generate new content and keep the site fresh, or people to promote the site, the appeal of Social Art faded. The energy required to maintain a place for artists to display their work was immense, and so the idea for Social Art began to seem pointless.

By September of 2002, I had spent two years learning how to program computers in order to further my understanding of Internet technologies. I set out to learn the full gamut of web technologies and how they were interrelated. I learned many things including: database design and implementation, both object-oriented and procedural programming styles and how to design enterprise systems, how to seamlessly utilize applications with databases, how to administer networks and the servers on them, how to enforce security and optimize an application’s performance, and that the more you know about the server your application runs on, the more options you will have with your application. Throughout this time of discovery, I often entertained the idea of combining these skills into an art project.

I was introduced to a few articles written by net artist Mark Napier and was delighted to learn that other artists were combining art with technology. When I saw Napier’s java code program that created a dynamic image based on user feedback, I was inspired. I realized that I could use technology to create a dynamic image that could be printed on a larger scale. I had already learned how to perform loss-less manipulation with gif images, so an effortless way to create an image was just what I needed in order to keep SocialArt.com fresh.

SocialArt.com was reborn and I began to experiment once again with art on the Internet. I modified my original vision for an Internet art gallery. Instead of showcasing work by upcoming artists, I would showcase dynamically created art. I knew right away that the best way to generate dynamic images would be with the use of keywords. I was very familiar with the way Internet search engines functioned, and knew that I could use them to get the locations of relevant, downloadable images. This process was the perfect source for my art. Using the search engines I could provide users with the ability to create art with words.

I decided to kick-off project development with a catchy name. Inspired by Napier’s talk of art machines, the name “art machine” seemed appropriate. I later changed the name to “the artists’ studio” because it seemed to fit better with the theme of the website. Regardless of the name, I saw this project as the quintessential art machine. Like art, it would be pretty to look at, and like a machine, it took input and provided output. I quickly moved on to creating the initial proto-type. I knew if I could figure out how to download the appropriate images based on the user input, and find software to collage them, I could finish the Social Art Project. Within three weeks I had what I needed to feel confident as I moved forward with my second attempt at an Internet art gallery. I had an application that resembled a search engine, but instead of returning a list of pages containing the searched keyword, it returned a single collaged image. If I wanted an image of monkeys, I would type in monkeys and have the search engines retrieve the locations of monkey images. My application would then download and collage the images.

It took me several attempts to get the right formula for the placement and blending of the images. My original plan had the images randomly placed on the canvas with a random rotation, but eventually I settled on a random placement without any rotation. The software I chose to collage the images with provided many of the same options as Photoshop. Filters are one example of this. I started out with one filter option, difference, and was pleased for the first week. Eventually I grew bored with the tie-dye effect caused by difference, and expanded the filter options to nine.

Several friends assisted me with the development of this project by using the application to generate images based on their keywords. This allowed me to get a better understanding of how the application would perform with increased usage. I had decided early on to save each image created with the application. I did not anticipate, however, that I needed to save each image based on a unique ID rather than by the query string. It was a rookie mistake, and I lost several hundred images because of it. To correct this situation, I decided on making a database-driven art gallery application capable of indexing and storing every image created with the application.

I tried to model my work after traditional concepts used by museum curators. I knew I would eventually have a large body of art work, perhaps eventually the largest in the world, so I knew I needed a flexible way of storing and retrieving each image. Like any inventoried item, I chose to use a unique number to identify each image rather than the name. This helped me save every image, and with a relational database it was a snap to store other vital information such as the query string that created it, and the time and date it was created. I found the search phrases my friends had used very interesting, and began to realize that the real appeal of this project would be the words that people searched for, rather than the tool that created the images. Although the tool helps, I now believe that if the tool is any good its product will be better. I imagine that this statement should hold true to all “art machines”, or at least should be every artist’s goal in making one.

The way the images looked concerned me most. If the finished product, the image, lacked intrigue, the success of this project and the usefulness of an Internet art gallery would be minimal. If I have learned anything using Photoshop, it is that with filters you can usually make any image look interesting. And being generated dynamically by the computer, no part of the image will ever be contrived. Manipulated images, with programs like Photoshop, usually look contrived because the artists went overboard somewhere on the canvas. With no bias, the computer-generated images are truly unique.

There have been many things throughout this project that have made this experience rewarding. For instance, the images that are downloaded to make the collage are significant. Typically, search engines are based on popularity. The images you receive based on your search phrase represent the popularity of a group of images during a specific time period for a specific search engine. As time progresses, different images move up and down in popularity. For instance, if you type in “sex”, the first image you get is that of a popular TV show called “Sex and the City”. Two years ago you would have gotten some porn image, and tomorrow you may get something completely different. The key here is that the ten or so images you receive for your collage are the ones most commonly accessed by the general population. In short, the general population has a major role in the outcome of your collage. The name Social Art now seems more fitting.

With an endless amount of phrases to search with and images to pool from, the potential for dynamic artwork is endless. When working with the Internet, however, it easy to forget the limitations of this medium. Bandwidth, for example, is the single most challenging aspect of the Internet. Bandwidth played a decisive role in the outcome of this project. My original plan was to amass a pool of thousands of images and create a single collaged image. In my initial test, I tried only twenty images and found out that in order to download twenty images it took over a minute. Then the images had to be collaged together, which took several more minutes. Cleary, in order to create a successful Internet project speed is key. Society has a high demand for high-speed Internet, and becomes impatient waiting for content to load. I knew I must decrease the time it took to deliver an image if this project would ever be successful. I decided to limit the number of images that were processed to ten. I now had a manageable completion time of around a minute. Surprisingly, fewer images looked better because there was less clutter. Everything seemed fine until, once again, bandwidth caused me to make another unfortunate decision. There was not enough RAM or CPU available to handle multiple processes from multiple persons. To counteract this, I had to create a waiting room application that managed how many people could use the application at one time. Hopefully, as time progresses, I will be able to find a way to serve this application to thousands of people simultaneously without a waiting room or any bandwidth restrictions.

The Social Art Internet Art Gallery may not be around in a few years, but the pictures that were created by it and printed will surely last several generations. My goal, with this project, was to create something with technology that had a life longer than the technology that had created it, and to create something that was constantly evolving based on user interaction. This goal has been met. Not only do I have a self-updating gallery with great images, I have the ability to scale each 700x400 pixel dynamically generated image into a print ready 46x64 inch image. Not all of the images that have been produced by the application have been masterpieces, but more than less are appealing. I am very excited to see how many images will be created with the application, and am even more excited to view them. I have laid the framework for an evolving project and must now maintain it. In programming, things usually need revision. As users interact online and provide feedback, I will attempt to make the necessary adjustments in order to fulfill the demand. I hope that since this application is dynamic the commitment of my time will be minimal so that I can spend my time enjoying the final product.

Shawn Moore
shawn@socialart.com