Kirk Woolford: Reckless Eyes

This project is supported by:

"As soon as I see, it is necessary that the vision… be doubled with a complementary vision or with another vision: myself seen from without, such as another would see me, installed in the midst of the visible…"

Maurice Merleau-Ponty, "The Visible and the Invisible"

Reckless Eyes video


Reckless Eyes used a combination of mobile wireless networks, handheld computers, video streaming, and 2.4Ghz video to allow people to see themselves through other's eyes and the eyes of their environment. The project focused on how people gaze and are gazed upon in an urban environment. Unfortunately, the technologies used in the project proved very unstable. Development took far longer than expected and ate a great deal of money and energy. Several weeks of tests culminated in a working experiment in early April, but during the final presentation on April 7th, two cameras died and the entire network collapsed.

However, when the system worked, it generated interesting and unexpected results. The wireless cameras received visually interesting interference from the environment, and the buffering from the streams created lags allowing participants to be in the same location seeing each other's gaze offset in time.

Conceptual Background

"Reckless Eyeballing" is a concept familiar to most Americans. It is most infamous for its use by the pre-civil rights courts to punish and imprison black men for looking at white women. However, it is still used for any form aggressive or inappropriate gaze, especially when a person without power dares to look a person with power directly in the eye. American prisoners are frequently thrown into solitary confinement for staring their guards in the eye, and prostitutes are in danger any time they look a pimp in the eye.

The concept of gaze has taken a radical shift over the past 50 years. With the advent of the CCD, surveillance cameras have spread numerous gazes across our cityscapes. The gaze of technology is different than the gaze of biology. The technical gaze can hold, record, and re-present the images before it. Most importantly, the technical gaze can be shared - either through re-presentation of the recorded images, or through live transmission.

From 1994-1996, Steve Mann, the grandfather of wearable computers, wore a wireless camera and receiver for almost every waking minute of his life (he took them off only swim, shower, and sleep). Both the camera and display were connected to the Internet so visitors to Steve's www site could see what he was gazing upon, and if a visitor sent him an email, it would pop up in the display before his eyes. Steve tells jokes about viewers to his website telling him to say hello to people the recognized or chastising him for "ogling cleavage".

Steve's experience of a shared technical gaze explored the concept of sharing a live gaze with others - with allowing others to view your world through your eyes. "Reckless Eyes" was the first step in an attempt to show people themselves as viewed by others and as viewed by the environment itself. It is highly influenced by Maurice Merleau-Ponty's writings in which he frequently refers to the seeing and the seen. Merleau-Ponty fights the Cartesian model of vision where individuals see the world from an external vantage point, or "God's eye" view and stresses that we as human beings are not dis-embodied eyes looking down upon the world. We are embodied. Therefore, in order to see, we must be in our bodies, in the world. If we are embodied in a shared world, we can be also be seen by those we see. Merleau-Ponty identifies the fundamental "reversibility" in vision: the observer is both subject and object, the seeing and the seen.

Merleau-Ponty's views on, erm, viewing put Steve Mann's experiment in a different light. Steve, attempts to give others a god's eye view of the world. Albeit, he attempts to make himself into a god by showing other's the world through his eyes. Steve is a benevolent god though. He allows his followers to speak to his eyes. Other would-be gods do not want to share their vision. In his writings on the Panopticon, Jeremy Bentham attempts to break this reversible vision through architectural strategies. He describes the first methods of mediated sight where the body of an observer is obscured behind a technology through which he can look out. However, when the observed look back, they see only the mediating technology, not the observer. Bentham tries to break the bi-directional gaze.

Technologies for mediation of gaze, have made radical leaps since Bentham's time. The advent of the CCD and surveillance cameras have spread numerous gazes across our cityscapes. A CCD's gaze can place the observer further from the observed than Bentham ever dreamed. However, the technical mediator (or camera) must remain in the proximity of the observed. Therefore, if the observer wishes to move away from their gaze mediator, they need some form of communication. In other words, a distant observer can view through a remote camera, but the signal from the camera needs to be transmitted to the observer. Traditionally, this is done with a direct wire between the camera and a monitor. However in recent years, it has become easier and more cost effective to connect the camera to the recorder via wireless radio transmitters.

These transmitters open a new concept in reckless eyeballing. Anybody with a proper receiver can pick up the transmission between the camera and the recorder. However, according to current Dutch Law, this is illegal, even if the camera is pointed into public space and the signal is intercepted in the same space. In other words, it is illegal for a person to look at how they are looked at. It is illegal for them to gaze into another person's technically enhanced eyes.

Reckless Eyes allows people to view through each other's eyes. It specifically seeks to show people themselves as viewed by others. It looks at manners in which concepts of sight and gaze are altered when they are technically enhanced, when sight can be passed beyond the individual.

Phase 1: Cross-eyed

The "Reckless Eyes" project is too ambitious to be built within the schedule and budgetary limitations of the Playing Fields project. We decided to use Playing Fields to focus on a single aspect of the overall project. The first phase of the project was called "Cross-eyed" and dealt with viewing oneself through another's eyes, or, more specifically, viewing yourself through a camera strapped to another person's head.

Phase 1 was intended to be a technical phase working out what we mistakenly thought would be a simple process of transmitting images from both people's cameras to a central location, digitizing the signals, and sending them to a streaming server. In addition to the head-mounted camera, each participant was given a handheld computer or PDA. The PDAs used a wireless network (802.11b WiFi) to connect to the streaming server and watch the video stream. This seemed like a simple process, but it took nearly 4 months to get our first mobile stream working. Some of the problems we had to address were:

- Camera transmissions and ranges: We chose to work with the 2.4 Ghz network because it is the most open network for video transmission and we knew of several cameras transmitting at 2.4 Ghz in the area around the Waag. However a licence is needed to use a 2.4 Ghz transmitter more powerful than 10 mW. A 10 mW transmitter has a maximum range of approx 100m. We spoke with icenced radio operators who said we could obtain 1W transmitters with a 1Km range, but they are powerful microwave transmitters and are dangerous to place near the human body. They also require large power sources and were outside our budget. We ended up using two "CCD Finger Camera"s from Conrad Gmbh, and two 10mW "Airlink Wireless" 2.4 Ghz transmitters from Maplin UK. The 2.4 Ghz video receivers were TranW "GigaAir TTA-10R"s also from Maplin UK.

- Choice of PDA media player and streaming server: The choice of PDA, client, and server are all tightly interconnected. It took several months of research to decide which system to use.

o We wanted to use open-source systems as much as possible so we originally explored Apple's Darwin streaming server with Sharp's Linux based PDA called the Zaurus SL5500. At the time, there were several MP4 players available for the Zaurus, but none of them could connect to an RTSP stream. After searching the net and testing several players, we decided to write our own java-based player for the Zaurus. Unfortunately, the Sharp Zaurus, and most other PDAs use an obscure version of Java called "Personal Java". Sun has already phased out this specification and made it an extension of their Java2 Micro Edition (J2ME) standard. The Java Virtual Machine (the program which runs Java on the PDA), is a version called Jeode. Interestingly, Jeode is the same VM used for most of the PocketPC PDAs. If we wrote a java-based player that ran on the Zaurus, it should run on the PocketPC as well. We were able to write a java player to connect to a stream using the "Java Media Framework", but the player would not run on the Zaurus. We finally discovered some stripped-down streaming video classes intended for PDAs and mobile phones called "J2ME Mobile Media", but never tested them on the Zaurus.

o The most common PDAs on the market run Microsoft's PocketPC2002 (formerly called WindowsCE) or PalmOS. Of the PalmOS devices, Sony's $800 Cliè was the only one for which we could find a WiFi adapter. HP, Toshiba, and Siemens all released PDAs with built-in WiFi while we were working on the project, but they all cost around €750.00 each - which put them out of our budget. There were numerous PocketPC PDAs on the market for around €250. Unfortunately, none of the inexpensive PDAs from Compaq, HP, Toshiba, or Viewsonic had a compact flash slot or WiFi network adapter. However, we learned that Compaq sells 'expansion sleeves' for it's iPaqs. These sleeves add extra batteries, Bluetooth, CF, and/or PCMCIA card expansion.

o We planned to purchase 2 Dell Axims, but managed to get the Brighton UK Based games testing and localization company, "Babel Media" to loan us an old Compaq iPaq 3760 and an iPaq 3850. We found two CF expansion sleeves and CF WiFi cards for these in London and finally had working mobile media players.

o The iPaqs run Microsoft PocketPC 2000 so they have a customized version of Windows Media Player bundled into the OS. Montevideo didn't have a Window's 2000 or XP Pro server and none of us wanted to work with Microsoft software, so we decided to look for an alternative to Windows Media Player.

o RealNetworks changed the name of their streaming client from "Real One" to "Helix". Under the Helix name, they made portions of the software open source. There was no version of the new Helix Player for PocketPC. But the "Mobile Real One" player works with most PocketPC based PDAs. Mobile Real One Player played files created with the new Helix Producer, but Helix Producer Basic (the free version) could not create a stream which could be played on the PDAs. We didn't want to buy an expensive commercial licence for the full version of Helix Producer, so after a fair deal of research and experimentation, we learned the Mobile Real One Player only played Real Media 8 streams. We dug out a two-year-old copy of Real Producer 8.5 Basic, and were finally able to stream to the PDAs.

- WiFi network ranges: One of the central themes of the project is bringing the gaze into an urban space. This meant we needed a network which extended into an urban environment. The Society for Old and New Media (Waag Society) is the Internet-2 backbone for Amsterdam. Montevideo's streaming servers are housed at the Waag. The Waag also runs one of the only open WiFi networks in Amsterdam. It seemed a natural decision to use the Waag's network for the project. However, the Waag is a less than ideal building for a hosting a network. Because it is a historical landmark, the WiFi antennae is not allowed to reach above the building's roof. The roof itself is made of nickel causing the WiFi signal to bounce erratically. The network coverage around the building has mysterious strong and weak pockets because of the shape of the building and the roof. In the end, we had to purchase our own wireless access point (D-Link DWL900-AP+ WAP) and place it in one of the windows of the Waag. We tried to use the D-LinkWAP as a repeater or bridge to extend the range of the Waag's network. Eventually, we realized that the D-Link WAP could not act as repeater for the Waag's network, so we stopped trying to use the Waag network and used our own.

- CF WiFi Cards: We originally used 2 Buffalo 128b CF cards with the iPaqs. These cards worked, but had very limited range. After we purchased our own WAP, we realized one of the Buffalo cards had been damaged during our experiments. We needed to replace the Buffalo card quickly but the only CF WiFi card we could find was a Netgear MA701. The Netgear card turned out to be a much better card with at least 25% more range and far more helpful drivers and software including ping, and traceroute.

- WiFi interference: We used 2.4Ghz transmitters and receivers because we wanted stray external video signals to interfere with our transmitters. Unfortunately, WiFi networks also work on the 2.4 Ghz range and created a fair deal of interference visible as white horizontal lines scrolling through the video feeds. We placed the video receivers and the wireless access point as far away from each other as possible to minimize the interference, but we could not eliminate it altogether.

- Delicate Hardware: From the very beginning of the project, we had problems with hardware breakage. We lost one transmitter and one of the WiFi cards during early development. On the day of the final presentation, one more camera died. As the audience entered the presentation room, the backup wireless camera and transmitter died as well. We assume the first transmitter was damaged because it was not properly encased and moved too violently when it was carried. We solved this by wrapping the transmitters in layers of foam rubber. The CF WiFi card probably broke through similar rough handling. We do not know why the two cameras died, but it may have had to do with the fact that we did not have any form of power filtering between the cameras and the batteries.

- Web browser windows: Once we had two live streams running, we tried to watch the two streams side-by-side. This is not a problem with Quicktime, but we moticed the RealNetworks Helix Player would not let us display 2 windows simultaneously. We suspect we can find a way around this limitation, but ran out of time looking for a solution.

Final Conclusions

While it can be debated whether the project was successful or not, it did yield a single fully-functional test. The aesthetic of this stream was far different than I expected. In my original concept, I forgot about the buffering that occurs in video streams. Once we had live streams running we realized the streams were offset in time. We also noticed the signal strength and CPU speed of the PDA affected the amount of buffering of the stream and the amount of time-lag between transmission and reception. We knew from the beginning that the signal from the wireless cameras would fluctuate. In fact, we used 2.4 Ghz specifically in the hopes of finding interesting interferences. However, the interference in final streams was again far different than expected. In addition to the image jumping and rolling, the streams would freeze, suddenly give bursts of several seconds of clear image with smooth, sharp images, then break up again. Instead of the look and feel of a television broadcast, the stream had a rough, broken, feel to it. I found this aesthetically interesting, but the project ended just as it became interesting.

Unfortunately the project suffered immensely from lack of proper funding. All of the wireless problems could have been solved by purchasing a couple PDAs with built-in WiFi support and purchasing 2 D-Link WAPs. We lost a great deal of time testing different WiFi cards and trying to find solutions which could work together. We could have developed and tested the system with our own network, then taken the network to the Waag. The issue of video transmission is a bit more difficult. With more money, we could have purchased more powerful transmitters and more rugged cameras. Ideally, we would use 900 Mhz video transmission to avoid the interference from the WiFi network. Another alternative would be to use portable computers instead of PDAs and digitize the video without any wireless transmission. However, both of these solutions would take away the possibility of wireless cameras in the area encroaching into our signal. Cleaning up the signal would sterilize the content.

Interview with Kirk Woolford

Biographical info:

Kirk Woolford is an independent designer, photographer, and programmer. He has worked both as technical and creative director for numerous online entertainment and education systems. In his own work, Kirk explores the use of technology to abstract the human body, its movements, and its senses. He was a research fellow at the Academy of Media Arts, Cologne from 1993-1995 and has won awards from ISEA, Ars Electronica, Arts Council of England, Ministry of Science and Education North Rhine Westphalia and others.