Contents 1 Underlying Issue 2 Concept 3 Design 4 Logistics 5 Potential ARG Use 6 Indoor Implementation 7 Interactive Possibilities

Underlying Issue

The transformation of the retail sector:

A series of projects have endeavored to analyze the Venetian retail sector in relationship to the comfort level of the residents throughout the city. Comfort level in this context refers to the ability of residents to purchase the basic necessities of life within a reasonable walking distance and at a reasonable price. What was discovered was that because of a decline in the population of Venice since WWII, the number of local shops stocking these has been dwindling. Once started, the cycle was hard to stop. It may have been that the population decline caused stores to close, but then the lack of markets within walking distance caused more people to move away. To maintain its vitality, the retail economy switched to the tourism trade, focusing on souvenirs. Stores offering necessary daily items have increased their price in order to afford the increased cost of building space.[1] [2] As this progression continued, some stores have remained empty and others have been replaced with modern-looking businesses. The unique charm and personality of the city is being interrupted.

Concept

The artistic vision of this installation is to “revive” some of the old businesses that once added charm and functionality to the city. This would be done by projecting images of the old storefronts with activity inside onto the empty windows of the old buildings. This would make it seem like the stores were active again, and would draw attention to the fact that they are not actually there anymore, but remind people what they were like while they were in business. Reminding people what was lost could help motivate them to preserve what is left, or restore what is missing.

The display would include the activity of people in the store, making purchases and interacting with the staff and each other. These would be silhouettes moving around the store. The image would be projected from either the inside or outside of the store onto screens set in the windows.

Design

In order to keep the display fresh and interesting, it is desirable to have the installation interact with the viewers. Whether the display is activated by a viewer is an independent issue from whether or not the video is pre-generated or real-time. To simplify, consider the four possible combinations below:

1. Video clips shown through the window are pre-recorded images and are simply displaying at all times
2. Video clips shown through the window are pre-recorded images and are displayed in reaction to the presence of a viewer
3. Video generated in real-time or time delayed using footage of real people in the street and are simply active at all times
4. Video generated in real-time or time delayed using footage of real people in the street and are displayed in reaction to the presence of a viewer

A system of displaying pre-generated video which is on at all times is the simplest to implement. Such a system would only require a projector and some sort of computer to supply the footage. The projector would be placed inside the store, making power and security less of a concern. This would back project on screens in the windows. A simple motion detector would activate the projection only when there was general activity in the street, to save on power and lamp life of the projector.

To cause the system to react to a viewer, a simple motion detector could be used. The basic application would be to have the system turn on whenever someone came within range of the installation. Another version of this would to be to locate the motion detector on another street, causing the installation to activate and surprise the person passing in front of the building.

Another way for the installation to react to a viewer, and also engage them, would be to limit the area on the windows upon which the video would be displayed. The image could be displayed only on sections that are shadowed by the observer, so the observer would need to walk up to the window and ‘look in’ to see the activity in the store. This method would require that the image be projected from the back of the screen. The system to implement this would also require a projector and some sort of computer to supply the footage. For the interactive possibilities, video cameras and some sort of motion and/or shadow detection device would also be needed. A typical data projector can be used to display the footage on the screen. The detection system can be done with the use of webcams, or standalone motion detectors. Motion detectors are available at typical hardware stores. A webcam would provide more functionality in one device, but it becomes more complicated to actually implement functions beyond a simple video capture device. There exist common software applications that allow the webcam to record video footage, and then to store, manipulate or display it. In order to use to a webcam as a motion or shadow detector, however, a custom software application would be needed to analyze video feed and determine the composition of the footage in terms of movement or shadows and their locations. A standalone motion detector could be incorporated without much difficulty to begin the display when sufficient movement is detected in the vicinity.

The simplest way to implement the effect of the shadow trigger is available only during the daytime. The sunlight will naturally drown out most of the visible image, making the screen appear mostly blank. When the observer approaches the screen to investigate, they would cast a shadow on the store front. This will reduce the interference from the sunlight, and allow the images to be visible. If the same effect were to be implemented at night, or when there is less sunlight to work with, the process would be much more difficult. This is partially due to the fact that less of the projection will flooded out, but moreover due to the reduction or lack of the shadows themselves. Even if this were to be implemented by actually detecting the shadows, the system would have more trouble in lower light situations. A more effective method might be to detect the location of the person, and match their outline with the projection area. This would only require that there is enough light to locate the person. The computer that controls the projector would need to know where on the image the observer or their shadow is, and only project in that region of the screen. This would require devices to identify the location of the person or their shadow outside the building, and have a method for the computer to isolate that region of the image to send to the projector. See Lozano-Hemmer for more details on shadow use in projection visibility.

The activity in the store window could be somehow derived from video footage of a nearby street. This method would very subtly engage passersby and observers by incorporating them into the installation. This involvement could be real-time, time delayed, or just loosely based on the footage.

In order to incorporate footage from the streets into the projections, there must be a way to process the video feed from the cameras. If the street footage is to be incorporated directly from the feed into the projection, then all that is necessary is to have the videos combined in the desired way. If the video feed is to be modified before incorporation, more advanced video processing will be required. In order for the people on the street to be used as a basis for silhouettes, their bodies must be identified within the video footage. This could be roughly done with a camera that has a manually adjustable focus. This would allow it to be setup such that the focus is either very far away from the storefront, or focused in the near field. This would cause the images of the people to be differentiated. If the camera is focused far off, then the people in the image would be very blurry. If the camera is focused near to the storefront, in the near field, then the people will be in focus, and the background will be hazy. Either of these methods could be used to aid the computer in isolating the people in the image. Software does exist that can perform this function, but it expensive and complex to use. Once the people were identified in the footage, their outlines would then be extracted from the feed to be used in the projection.

Potential Equipment List:

• Projector for output
• Computer of some sort to perform the image processing and display
• Motion detector to trigger the projection
• Webcam or other image capture device

A possible projector for this installation is the ViewSonic PJ503D. This was chosen for its relatively low cost, and reasonable intensity. A possibility for the processing core is the Mac Mini, for its small size, easy portability, and minimum need for peripherals one established.

See equipment selection page for comparisons.

Logistics

1. Need to find shuttered shop
2. Need to find pictures
3. Matching picture + shop would be ideal
4. May need to start crowdsourcing web site
5. Find friend with shutterd store across the street
6. Might need a permit

Potential ARG Use

The picture could be slightly doctored to give "clues"

Indoor Implementation

For a different implementation of this idea, an installation could be set up in a display area such as the Venice Biennale. An image of the entire storefront could be projected onto a screen or a wall. This indoor display allows an extra degree of detail in the video because there is more control over the brightness of the area. In this way the "people" in the store can be more than just shadows. As an alternative, within the gallery, a stage type façade could be built for the installation as the base upon which the projection is made. This adds a convincing physical element to the display but also adds permanency to the scene. The projection could include videos of real people or animated images doing business in the store.

Interactive Possibilities

The interactive possibilities for the indoor implementation are similar. Cameras, video or still, could be placed in hidden or visible places to take pictures or videos of people who are visiting the exhibit. These images could be used to directly place the viewer in the store real-time, or could be stored for later. Stored images could be processed and used to add the person into the pre-generated video, to make them part of the life scene in the store. This could be done with face recognition software that would place the face of the person onto a template character. With more camera angles, image processing, and templates, a person could be fully incorporated into the scene. Over time with either of these methods, the original 'stock' characters in the store could be completely replaced with people who have visited the exhibit, possibly creating a database from which to pull a set of characters that rotates day to day, or even over the course of the day.

Return to Postmortem Main Page