The Technology Behind RockstAR


We recently had the opportunity to debut the RockstAR experience at SXSW – check out video of the experience in action. We like to think of it as the classic photo booth taken to the next level with augmented reality, multi-touch and social integration. Let’s go behind-the-scenes and take a look at both the software and hardware that brings this experience to life.

First, let’s talk software. The application was built on the recently announced Razorfish Vision Framework. The framework provides a platform to power augmented reality, gestural and other vision-based experiences. For the RockstAR experience, we are analyzing each frame coming from an infrared camera to determine if faces are found in the crowd. Once a face is detected, it is assigned a unique ID and tracked. Once we receive a lock on the face, we can pass position and size information to the experience where we can augment animations and graphics on top of the color camera feed. This technology has a variety of uses. For instance, face tracking can be used to track impressions on static or interactive digital experiences in the retail environment. Here is a screenshot taken from the debug-mode of the experience which shows the face tracking engine at work using the infrared camera.

In addition to the vision-based technology, the experience was fully multi-touch enabled – users can gesture on a virtual joystick to swap out bands and snap pictures.

Because the classic photo booth experience is a social activity, we took it to the next level with twitterand Flickr integration. As pictures were snapped, we’d immediately make them available online. A QR code was rendered with each picture to quickly allow users to navigate to the RockstAR photo on their mobile device. Once the experience is extended to mobile, users can email the pictures to their friends, set it as wallpaper, re-tweet it to their twitter followers, etc.

Let’s move on to hardware. Unfortunately, you can’t purchase infrared AR-ready cameras at your local Walmart… at least not until Project Natal comes out later this year. Therefore, we needed to build a dual-camera system that would support the face tracking in infrared and the color video feed for display on the screen. We decided to go with 2 commercial-grade Firefly MV cameras with custom lenses.

One of the cameras we modified to see only infrared light by replacing the IR-blocking filter with a IR band-pass filter. This allows only a narrow range of infrared light to reach the camera CCD.

We also purchased and tested a variety of infrared illuminators. These are used to illuminate the environment with invisible infrared light allowing the infrared camera to accurately track faces in low-light conditions.

Sparks were flying as we fused the color and infrared cameras together — just another day at the office.

We created a portable rig for the camera and infrared illuminators. Adjustable camera mounts and industrial strength velcro provide flexibility and portability across a variety of installations.

We used a presentation remote clicker as an alternative way to drive the experience. We primarily used it as a remote camera trigger which allowed us to quickly snap pictures of unsuspecting people from a distance.

The experience was powered by a 55″ multi-touch screen and a CPU provided by DFI Technologies. We’ve been working with DFI to build PCs that will power the next-generation of interactive experiences. These PCs are small form factor and can be mounted behind the multi-touch screen.

Comment