Art from code - Generator.x
Generator.x is a conference and exhibition examining the current role of software and generative strategies in art and design. [Read more...]
Archive for February, 2006
Matthew Lewis: Sketch

Matthew Lewis: Sketch Left: Complete image. Right: Detail. More sketches here.

Who says idle browsing of Flickr images can’t be productive? While using Doug Marttila's findr hack to browse tag structures on Flickr, this lovely Sketch popped up.

Some background research reveals another page with little more info. The piece is by Matthew Lewis, a “Graphics Research Specialist” at Advanced Computing Center for the Arts and Design (ACCAD), Ohio State University (OSU). Sketch is from 1998, and the technical platform used was AL, developed at ACCAD by Stephen May. AL is a Scheme-based environment for procedural computer animation, and is still available for Linux platforms.

Lewis calls the piece a drawing machine. In his own words:

Inspired by [Harold] Cohen, I’d written some Scheme software to represent and generate shapes consisting of clusters of “marks”. The program generated postscript that was then printed.

The theme of automated drawing crops up again and again. Drawing machines practically constitute a sub-genre of generative art. Lewis’ machine is striking because of its whimsical, hand-drawn quality. It avoids a typical computer-drawn look, and manages instead to capture a more naive form of line drawing.

Matt has been good enough to make more Sketch images available as a set on Flickr. They demonstrate a consistent quality of line, somehow cartoon-like. Click the “All sizes” button to get up close and personal with a high-res version.

Lewis has also created a body of generative work done in Jitter, including some work with genetic algorithms. He also also teaches Jitter at ACCAD. In 2004 he and Hans Dehlinger (one of the original Algorists) presented a paper at the Generative Art Conference in Milan on generative line drawing using photographs. Have a look at his Generated set on Flickr for more images.

Joshua Davis: Z4byJD

Joshua Davis: Z4byJD. Left: The prints (details). Right: The car, the artist.

Joshua Davis is at it again. This time he was commissioned by BMW to do a series of prints based on their new Z4 Coupe, in what could be considered a followup to the BMW Art Cars. The result is Z4byJD, consisting of a limited series of unique generative prints and a rockstar-style web site documenting the process.

The Z4byJD site, while highly polished and well executed, verges on pure designer camp. The site’s intro sequence begins with the payoff “The most radical artist… takes on the most thrilling BMW”, set to a rock/hip hop soundtrack. Davis is shown posing thoughtfully kneeling by the car in a clean showroom, inspecting every line of the car while the camera makes much of his famous tattoos.

Once past this curious opening, visitors can enjoy some generative graphics courtesy of Davis, and a “making of” video that threatens to become even worse than the intro, but miraculously saves itself at the last minute. After a frenzied start, it calms down and has Davis speaking with the soft-spoken Chief Designer Adrian van Hooydonk. The rock guitars give way to pleasant ambient music and talk about lines, light and surfaces. Joshua Davis explains his generative work processes, and the BMW heads make a comeback explaining their support for experimental art.

So despite Davis’ coarse initial quip about the car’s “sexy ass”, the video suddenly becomes plausible high-class BMW branding. This is a luxury car they’re selling, after all.

The prints (shown in a Gallery section with curiously low resolution images) do not literally reference the Z4 Coupe, even if segments of the car are buried in the composition. The images follow the vein of the organic-feeling generative systems Davis show on Once-upon-a-forest, a look that he does very well. The work has an effortless organic quality to it, with great colors and composition. A single print (all unique) can be bought online for EUR 275, not a bad price for a 61 x 112 cm print (unmounted). Obviously, the selling of the prints is not the money-making part of the deal.

It’s interesting to see that BMW (or at least their ad company) think that promoting a new media designer slash electronic artist in this way will help them sell cars. Using artists to endorse luxury products is of course common-place, and the BMW Art Cars do set the precedence, but still this seems an odd pairing. Then again, maybe Joshua Davis really is the Eminem of new media.

Just for balance, here’s some alternative takes on the same project: MPH Magazine, Autoblog, Dexigner, Leftlane News. Make up your own mind, don’t listen to us.

Takeshi Murata: SHIFT cover, issue 085

Takeshi Murata: SHIFT cover, issue 085 (1 left), Melter 02 (2 right)

The veteran Japanese webmag SHIFT has impeccable taste in cover designs. Every issue has a commissioned piece by a designer or artist, who is interviewed for the magazine. Typically, they focus on digital media or motion graphics, so the covers are usually web-friendly animations or interactive pieces.

While searching for images by Takeshi Murata, I just stumbled over this gem: SHIFT cover, issue 085. It’s from December 2003, and it’s lovely. Murata made the cover in connection with his animation Melter 02 being selected for the .MOV festival in Tokyo, using the same (non-programmed) technique.

Based in LA, Murata creates “experimental and psychedelic animations” by his own description, somewhere between the disciplines of art and design. This kind of hybrid approach is typical of the LA motion graphics crowd, with some spectacular results. Designers like Geoff McFetridge, Logan, Geoff Kaplan etc. experiment with personal work and use it to develop eclectic styles. (Which then gets them hired by big companies, so it’s not all love and flowers…) is a little thin on info, so be sure to read the SHIFT interview with Takeshi Murata and watch the Melter 02 clip.


Despite our enthusiasm for folksonomy and tag-based navigation, the Generator.x blog has not utilized tagging to manage its own content. While all entries were tagged, those tags could not be used to navigate the blog itself. Until today. Help has arrived in the form of Christine Davis' nifty WordPress plugin Ultimate Tag Warrior 3.0. As the name would indicate, this plugin presents a complete solutions for all things tag-related.

As a result, there are a few new features, although they can be hard to spot right away. Where previously the tag links under the post titles would lead to a search on Technorati, they will now present you with an archive of all entries tagged with that keyword. You can also access the tag search directly with URLs of the form, or choose Browse Generator.x by tags from the Categories menu.

There is also a small collection of the most popular tags in the sidebar. It doesn’t look so great right now, but it provides the basic functionality. The Generator.x blog templates are due for an overhaul in the not too distant future, the tag list will be more elegantly resolved then. For now, the new tag features add to the ease-of-use of navigation.

A behind-the-scenes change is the migration from tracking visitors using, to installing the excellent Mint stats package. At $30, it’s hardly much of an investment, and the design and presentation of information is spot-on. In fact, one could argue that Mint features some of the best table design on the, for proof, see the demo over on Add to that the fact that like WordPress, Mint is specifically built to be extended upon and there are already a host of great plugins, it was an easy choice.

Name: Project

Amber Frid-Jimenez and Philip DeCamp: Shrub: Variations on tree structures

Shrub is a series of sketches of tree structures, by MIT Media Lab students Amber Frid-Jimenez and Philip DeCamp. It starts off looking as one would expect, then wanders off into the exotic. Some feel more like drawing than visualizations.

Both Frid-Jimenez and DeCamp were involved in another project called Document Icons, a software sketch designed to allow the user to search through the contents of millions of text documents using histograms. The technique looks like it could also work well for tag heavy sites. Infosthetics blogged it here.

(via and infosthetics)., the brain-child of UK artist Stanza and veteran survivor of online interactive art, has relaunched in new glory. Retooled by Karsten Schmidt (aka toxi) in a new “web2.006 framework” called @emitter, Soundtoys has made the jump overnight from old-school semi-static site to being all-dynamic, tagged and blogged.

Soundtoys contains a wealth of old and not-so-old works, including more than a few gems from the productive period of Shockwave experimentation in the mid-to-late 90s. To get a good idea of the scope of the project, have a look at the artists page. It reads almost like a who’s who of a cross-section of new media art and experimental design.

Adam Hoyle and Julian Baker have contributed a content navigator, which allows interactive navigation of the project database using the tags. Users can combine individual tags to combine them as a search, get a short description of the piece and launch it directly from the navigator. Although some of the pieces may make your browser complain about Shockwave’s brutish ways (or even crash outright), persisent browsers can expect (re-)discovery of particular gems.

Many props to Stanza for starting this community and keeping it alive, and to Karsten for his good-looking but ├╝ber-functional publishing platform.

Update: Karsten just blogged a detailed description of the magic behind the scenes of the new Read it over on

Some highlights:

Leonardo Solaas: Dreamlines

Leonardo Solaas: Dreamlines

Leonardo Solaas: Dreamlines

Leonardo Solaas: Dreamlines

Argentinian generative artist Leonardo Solaas has created a dreamy, painterly work in his Dreamlines. The user inputs one or more keywords as input, which set the piece off on a a Google image search. The resulting sequence of images are then used as the basis for an animated drawing, morphing and changing over time.

The drawing process itself is done by a swarm of particles, who navigate the image as though it were a virtual terrain. Formulas translate pixel values into instructions for particle movement, making them relatives of finite state machines. The results are surprisingly subtle, the quality of the stroke ranging from hairy to spiky, curvy etc.

The interim state when one image is replaced by another is the most interesting, as one image is erased and one can clearly see the lines drawn by the particles. Obviously, depending on the search the resulting drawing can be recognizable as a filtered image or not at all. One tip is to input your own name or a well-known brand name. The right-hand image shown above is a re-drawing of an image used for an interview on Since the original contains limited colors and sharply defined shapes, much of the form and color is retained.

Quote from the artist:

Dreamlines is a non-linear, interactive visual experience. The user enters one or more words that define the subject of a dream he would like to dream. The system looks in the Web for images related to those words, and takes them as input to generate an ambiguous painting, in perpetual change, where elements fuse into one another, in a process analogous to memory and free association.

Thus, the work is at the same time a study on population dynamics, or on the emergent behavior of a multitude of very simple autonomous agents.

Who is dreaming? The user, or the Internet itself? In a certain way, both. The program generates a personal moving picture, unique, unpredictable, and forever gone when it is finished, just like dreams. But that dream is made out of pieces taken form the subconscious of the whole net, gathered by some words of the user and the obscure logic of searching algorithms.

Dreamlines was shown at Transmediale.06 and recently won the IBM New Media Art award at the 19th Stuttgarter Filmwinter.


With its rich content and well-implemented tagging system, provides a tantalizing data set for would-be information visualizers. Fortunately, the open API allows developers full access to the functionality of the system.

To support the recently launched Processing hacks site I have written up a quick tutorial on how to access with Processing. The hack uses David Czarnecki’s delicious-java library. I also added a simple hack for outputting PostScript vector files.


This site went live while Generator.x was having a holiday, but it deserves a repost even though it’s a few weeks old:

The ever-productive gentlemen Tom Carden and Karsten Schmidt (Toxi) have launched Processinghacks, a user-contributed Wiki intended to provide the Processing community with documentation of advanced techniques.

Processinghacks nicely fills the gap left by the lack of tutorials on the Processing site, combined with the beginner focus of the built-in examples. While a lot of answers are available on the forums, they are sometimes out of date or hard to find. Processinghacks provides details on specialized techniques that are beyond the scope of the core Processing project, such as integrating Processing with Java or hacking the source code itself.

A big plus is that this effort is completely independent of Ben and Casey, which means that they can focus their energies on the core project of bringing Processing to version 1.0. For those who remember the debate brought up by Karsten a little while ago, this should set an example. Instead of just complaining about the state of things, people like Tom and Karsten are actively providing a service to the community.

Some highlights from Processinghacks:

Daniel Brown: Flowers

Grafitti Research Lab: LED Throwies

LED Throwies is a project from Grafitti Research Lab, a division of the EyeBeam OpenLab. It involves hooking LEDs up to a small battery and using a strong but miniature magnet to make it stick to metallic surfaces. As the name suggests, the typical mode of application is simply to throw them at the target.

“Throwie” is a reference to graffiti “throwups”, quick and dirty pieces usually done with a single layer of paint and an outline. The LED Throwies could point the way to a new form of urban street art, adding color and magic to the hood. Look at the video on the Grafitti Research Lab site for a nice preview of a Throwie “party”.

There are no detailed credits on the Grafitti Research Lab page, but there is reason to believe that Evan Roth aka fi5e (who created the Grafitti Analysis piece) is involved somehow. Since the OpenLab is dedicated to public domain R&D, there is a publicly available detailed recipe published on

More images on flickr/tags/ledthrowies.