Great Wall of Pi: Creating a high-tech multimedia solution for $2,000, not $10,000.
The play centers on individuals meeting in a chat room for recovering addicts, with several scenes taking place in their online space. Additionally, with characters meeting in more than one dozen different real-life settings, the media design could assist with the creation of those locations. The solutions offer a model for budget-conscious productions that uses interdisciplinary collaborations, theatrical design tools, and the open-source software PiWall.
When Set Designer Ryan deRoos presented her rough design at the first production meeting in early May, it included a 23-display video wall lining the cement back wall of the Cellar Theater. Depending on the scene and location, these displays would show different configurations of images and video. For example, if the entire stage was being used by one location--a flower shop--during a scene, then the displays showed one entire image of the shop's interior divided up so that each display showed only a part of that image. If the stage held two separate locations (stage right a church, stage left a train station, for example), then the screens could be divided so that some displayed one image and others displayed a different one. Or, if the scene was taking place in the virtual chat room, individual displays would show the chat icons for the characters as they logged on and off. In addition to this, a 24th "display" was added: a front projector that would work with the video wall in the play's final scene to create the effect of a waterfall cascading down and across the stage floor.
Initial research into video walls and pricing was not encouraging. Playhouse San Antonio does not have a large budget, and its Cellar Theater shows are the smaller, more cost-effective shows in its season (as opposed to the musicals that take place on the mainstage). The video walls found in airports, stadiums, or stores can run upwards of $10,000. Rear projection, a suggestion made early on, wasn't possible in this space (front projection obviously was not a viable option due to the low ceilings).
The ACTLab at the University of Texas at Austin is something of an incubator for collaborative problem-solving efforts. The author was reminded of her own experience in the ACTLab with founder Allucquere Roseanne Stone while mulling solutions for the video wall for the Water by the Spoonful production. In the ACTLab, students of all disciplines come together for a semester to focus discussion and creation around one topic. The ACTLab is housed in a former TV studio in the Radio/Television/Film department, where students have access to computers, software, projectors, cameras, lighting, and sound equipment, as well as teaching assistants willing to help with projects. Many of the students were from computer sciences and were interested in hacking, and the class encouraged unusual collaborations. It also encouraged students to teach themselves: If you didn't know how to do something, go and figure it out.
Perhaps seeking collaborators outside of the theatre world was the solution the team needed for the Water by the Spoonful production. The author contacted Dario Landazuri, a systems engineer, through gaming circles in Austin. Landazuri thought a free, open-source project called PiWall might be a good solution. PiWall allows users to set up video walls with any number, shape, and size displays using Raspberry Pi computers. It was developed as an affordable solution for a video wall in 2013 by Alex Goodyear and Colin Hogben. The software package is available for download online, and communities of users and programmers are willing to help first-timers. It requires some set up and programming time--building each computer, installing the operating system, and configuring the software to the specific video wall layout in question. Each display is connected to its own Raspberry Pi, and a "master" Pi in turn controls them all. Once set up, video files can be sent from the master computer to the wall, and each Pi divides up the image and displays only its portion of the whole.
The online community surrounding PiWall is one of its strengths. According to Alex Goodyear, the Water by the Spoonful production was the largest noncommercial PiWall he knew of at that time; PiWall was developing a commercial version and working on a few productions that had more than 300 displays.
The first step in considering PiWall as a solution was to run a small test. The test required purchasing three Raspberry Pi computers on Amazon (3 Raspberry Pi 2 Model B Project Boards, 1 GB RAM, 900 MHz Quad Core CPU plus Tontec cases, 8GB memory cards, power supplies and adapters) and setting up a three-display PiWall in Landazuri's office as a "proof of concept." The video files are played on the wall by executing shell scripts on the master computer. Combining theatrical knowledge and software with the open source PiWall enabled the team to troubleshoot control scripts during this test. QLab 3 was used to execute the shell scripts, since typing out an entire command at a prompt for every "go" given by a stage manager wasn't a feasible solution for an operator during a performance. QLab 3 provided the perfect solution because it allows a user to add a "script" cue rather than a video, sound, or other typical type of cue. The script cue in QLab 3 contains a tab where the user enters the script QLab needs to execute upon "go." All the shell scripts could then be executed without having to enter a command at a prompt. Once the Playhouse team saw that this solution could work, assembly began on the other 20 computers.
Each SD card was installed with the software found on PiWall's website, making them the "hard drives" for the Raspberry Pi computers. Each Pi computer was assigned a unique IP address within the software that in turn reflected the number assigned to its "tile." In order to tell each Pi where its particular tile was located, a grid was placed over a front elevation drawing of the wall, which determined the x and y coordinates for the top left corner of each display. This was not done using any method PiWall suggested and instead drew again upon standard theatrical knowledge. Scale drawings were created in Vectorworks based off the set designer's Sketchup model, and a grid of two-inch by two-inch squares was laid on top of a front elevation of the wall. The coordinates (0,0) became the top left corner of the wall itself, and so display #1, which was the closest to that corner, started at (2,2). These coordinates were then installed along with the rest of the software onto the Raspberry Pis, and each unit was carefully labeled on the outside.
INSTALLATION AND TESTING
The built components then moved into the Cellar Theater at Playhouse San Antonio. The team that assembled the wall and all of its components consisted of four people, including the author, Landazuri, Set Designer Ryan deRoos, and Technical Director Pat Smith. The team had to work collaboratively, essentially brainstorming how to best proceed with building and installing the wall. The displays were mounted 14 inches away from the back wall of the theatre to accommodate the Pis and their associated cables and power supplies. Each Pi was installed as close as possible to its display, and the space between the displays was covered in black duvetyne rather than a hard covering, allowing for as much access as possible for troubleshooting during tech rehearsals. The Ethernet switch was located within easy reach. Ethernet cable was cheaper than buying individual lengths, and the theatre already had large spools in stock, which saved money, though not time.
It's worth noting that the displays were all "found" displays, with the exception of two large rented ones. Several were computer monitors the Playhouse owned, while others were borrowed or sourced online. PiWall's code allows for this difference in displays--resolutions do not have to be identical across all screens. Issues did arise with some of the displays and several had to be swapped out with new ones during tech.
Because the Raspberry Pis had been "told" the starting point of each display, anytime one was replaced, the size had to remain consistent. A few Ethernet cables were replaced as well, but these were the only gear-related technical problems during the entire run of the show.
That said, the team had to work around a number of oddities while using an untried solution like this for performance. When working with software such as Isadora or QLab, in the event a problem occurs, one assumes an elegant solution has been developed within the software itself. PiWall was not designed to do exactly what it was being used for: running more than four dozen separate video cues that had to be called and executed at precise times--some displaying video while others were still images and some only on one display, while others took the entire wall. The team did not have that luxury here, so the solutions may or may not be elegant as a result. However, given that all problems arose between load in and opening night (a week and a half stretch), and the solutions all worked for this show, the team was, at least for this project, satisfied.
Once the wall was built and it was functional, attention turned to making this work for the production. One of the challenges the PiWall presented was that it did not allow for individual display control. PiWall was essentially, as Landazuri put it, "one big dumb screen." A mask in After Effects was used to be able to isolate displays within the wall--first creating an initial rough version just by using the elevation drawing, and then displaying it on the PiWall. The mask could then be edited in Photoshop, with the saved changes updated relatively quickly to the live version on the wall. In this way, the author edited the positions of each rectangle until anything shown on it would only display within its tile. When creating all video files to be used in the show, the author imported each of them into After Effects and resized and positioned them until they only appeared inside the desired tiles, leaving the rest of the image black. This created the illusion the author was sending one video file to one display, when in reality one large video was sent to ALL displays, and it had been edited to only show in the area corresponding to the display in which it needed to be shown.
Another challenge: PiWall only played video files (mp4s), and many files were still images of locations in the play. When a video playing on the PiWall reaches the end, it simply freezes on the last frame, so turning an image file into a three-second movie file meant that after three seconds, the still image would remain until the next cue was executed. The solution was to craft fade-in and fade-out animations for each individual cue, which played before and after the still image. These appeared to fade seamlessly into each other because each fade-in animation ended in the still image, and fade-out animations began in the still image. Fading to black was handled in the same way.
Other challenges did not result in solutions, elegant or otherwise. The most common problem was caused by attempting to run two cues simultaneously. This is something that happens frequently in theatre, normally without issues. PiWall can only handle one video file at a time. And while the team was careful to create video files that were just as long as they needed to be, they froze on the last frame and the video files themselves could be very short (except for still images).
The bigger problems happened with preshow, intermission, and the waterfall cue in the last scene--the three times in a performance where a longer video was used. If cue one was still playing when cue two was executed, the result was green static and noise on all of the displays. Because most cues were seconds long, this was usually fixed simply by waiting for all cues to finish running. With the longer ones, all cues had to be cleared with the needed ones reloaded. Careful operator training and conscious design of the length of all cues was the best way to avoid this problem.
Some displays were more problematic than others, though there wasn't an identifiable pattern. Usually, rebooting the Pis and cycling the power fixed any issues. One display would occasionally just show a traveling white bar moving from one side to the other. When this happened outside of a performance, the solution was to cycle the Pi's power and reboot, but without being able to physically go up to the wall itself, an operator would not be able to do this during a performance. In a future iteration of the PiWall, it will be important to consider how an operator could easily cycle the power to all displays quickly and simultaneously, perhaps from the light booth during a performance.
The biggest unresolvable issue concerned image resolution of the waterfall at the end. The video file was high quality, and the different display resolutions should not affect the PiWall, according to its own creators, but a couple of displays were very pixelated. The creators believed that it was due to the quality of the video, but the team didn't agree, thinking it was likely a problem with the network, or an issue with the resolution of this video file being too high for some of the displays. The team was unable to resolve the issue during the run of the show and would need to devote extra time to troubleshooting this issue in the future.
Focusing on the overall successes of this project, however, show that they outweigh the challenges. The main success--and the reason for pursuing this approach-was the cost effectiveness of the solution. (See expenses chart on page 22.)
Using a PiWall for this media design was an effective way to achieve the results wanted by the designers and director for significantly less money than a turnkey video wall setup. This project also demonstrated how smaller budgets can often lead to outside-of-the-box solutions and collaborations. The use of open-source solutions frequently means that communities of users are willing to help you achieve your goal, and that was certainly the case here. The creators of the PiWall were also keenly interested in the project. At the time of the production, it was by far the largest PiWall that they knew of (most used fewer than 10 displays). Because the team figured out how to use Qlab to run the cues, Qlab's creators were also made aware of and interested in this design. While inexpensive computers served as a way to solve a problem in the script, others are using them as sources of inspiration to make new art [see sidebar]. Landazuri ended up keeping a Pi for his own personal experimentation after Water by the Spoonful was open. Being able to play with these computers, and brainstorm what was possible, was inspirational.
Megan Reilly creates original lighting, media, and scenic designs for performance. Her professional work includes media designs in Minneapolis for Mixed Blood Theatre, Walking Shadow Theatre, New Century Theatre, and Open Eye Figure Theatre, and lighting designs for Live Action Set and Red Eye Theatre. In Texas, she has designed for Austin companies such as Salvage Vanguard Theatre, Rude Mechanicals, Trouble Puppet, Whirligig Productions, and Hidden Room, and several productions with Playhouse San Antonio. She is an assistant professor of performance design and digital media at Macalester College. At Macalester, Megan Reilly has begun a scenographic practice of designing multiple elements of one production. She also investigates "experience design, " the process of creating entire worlds and realities that call for the active engagement of audience/participants in performances involving new media, immersive theatre, and alternate reality games. Upcoming projects include Transmission, an augmented reality/immersive theatre piece in the Edinburgh Festival Fringe, Mary Poppins with Prescott Park Arts Festival in New Hampshire, and If My Feet Have Lost the Ground at the Puppeteers of America Festival in Saint Paul, MN.
Kong, James. 2016. "Firefly: A wearable LED device with show controllable system." Hackaday.io. October 17. https://hackaday.io/project/16377-firefly
Upton, Liz. 2013. "Tudor Theatre, 21st Century Technology." Raspberry Pi Blog, August 29. https://www.raspberrypi.org/blog/tudor-theatre-21st- centurytechnology/
PiWall can be found at: http://www.piwall.co.uk/
Raspberry Pi Computers in Artistic Creation
Although Water by the Spoonful was the first production to use the PiWall software in its design, other theatrical and dance productions have used Raspberry Pi computers in scenic and costume design. For example, in 2013, Shakespeare's Globe Theatre in London produced all three Henry VI plays. The production was a collaboration with The Space, a program from the BBC designed to provide artists assistance in expanding audiences digitally. The plays were staged outdoors and streamed for free from multiple viewpoints. Among the many cameras used to stream the plays were the "Throne Cams," tiny cameras filming from the throne at the center of the scenic design for each play. This camera was a Raspberry Pi camera module, attached to a Raspberry Pi computer. The camera was able to record, process, and encode the video. Because of its size, it was invisible to the audience but provided a perfect actor's view of the performance (Upton 2013).
Other projects have incorporated Raspberry Pi computers into costumes for dancers to wear, such as James Kong's Firefly multimedia dance performance. In Firefly, dancers' costumes were lined with LEDs to create a glow-in-the-dark effect during the performance. Each dancer had on a small "backpack" containing the Raspberry Pi and other necessary hardware. And, as with Water by the Spoonful, QLab was used to control the LEDs (Kong 2013).
Many such projects that incorporate technology are found in experimental performance and art installations. The tools needed to execute are inexpensive, which makes them accessible to artists that might not otherwise be able to create multimedia art. Burnish is a performance installation that premiered at the Venice Biennale in 2015. It was created by Erika Batdorf and Mark-David Hosale, who is credited with much of the Raspberry Pi programming and design. It is an interactive performance for one audience member inside a mini-pavilion.
As performance that responds directly to input from audiences becomes increasingly popular, creators will find themselves engaging in collaborations with those who can make the work possible.
Caption: Water by the Spoonful at Playhouse San Antonio, directed by Jim Mammarella. Photography by Siggi Ragnar
Caption: Pi wall programming. Photo: Alma E. Hernandez / Reprinted with permission of the San Antonio Express-News.
Caption: Front wall elevation of video wall, showing numbered displays (directly above), and tech rehearsal (top). Photos by Megan M. Reilly.
Caption: Diagram of the media system (left) and formatting video in After Effects (above)
Pi Wall Budget Qty Item Cost Total per unit cost 23 Raspberry Pi 2 Model B Project Board, $48.79 $1,122.17 1GB RAM, 900 MHz Quad-Core CPU 23 Tontec New Raspberry Pi 2 Case $6.99 $160.77 23 8GB microSDHC Class 4 Flash Memory $4.29 $98.67 Card 23 Micro USB Power Wall Supply Charger $7.99 $183.77 23 HDMI to DVI adapter cable $7.99 $183.77 Displays--amount spent total $472.91 Ethernet switch--used Playhouse's $0.00 inventory Ethernet cable--used Playhouse's $0.00 inventory TOTAL $2,222.06 COST:
|Printer friendly Cite/link Email Feedback|
|Author:||Reilly, Megan M.|
|Publication:||TD&T (Theatre Design & Technology)|
|Date:||Mar 22, 2017|
|Previous Article:||WEARABLE SOUND TECHNOLOGY: Making magic in Caryl Churchill's Fen.|
|Next Article:||Front of House in the Forefront: Consider the possibilities when the lobby becomes a 'third place,' where enriching things happen at any time.|