AR without the Markers
The Ford C-MAX augmented reality campaign ran for two weeks across ten shopping malls with over 125,000 curious shoppers interacting with Ford’s Grand C-MAX vehicle. A selection of on-screen icons allowed participants to explore various features of the car including 'Door Slide', 'Seating', 'Power Tail' and 'Spin' with a 'Minority Report' style gesture-controlled interface.

The brief demanded us to track users in the most natural way possible and allow them to ‘play’ with an interactive product advert. No markers or other symbols could be relied on. It also had to ignore people walking past while allowing precise control and selection from the interface.

Proof of Concept
Inition were approached by digital production company Grand Visual who asked if this was possible and thanks to our recent distribution agreement for the Panasonic D-Imager (a Microsoft Kinect-like depth or Z camera) we provided a simple proof-of-concept demo within one day which secured the project.

Over the next few weeks, our in house team of C++ developers developed an application based on OpenFrameWorks and OpenCV SDKs to code the real-time sensing and video display engine. A sophisticated blob tracking and depth thresholding algorithm sensed the participant's hands allowing the car to follow the user's movements. A flocking algorithm gave a fluid, graceful movement around the screen with the optimum amount of 'bounce'. Full HD resolution, portrait graphics, webcam capture, depth and video camera aspect matching and alignment, gestural user interface, interaction logging for usage analysis and hardware fault tolerance were added before going live.

Fine Tuning and Feedback at Every Stage
An important part of the project was the ability to "fine tune" and "play" with all the control parameters via an on-screen interface. On-screen debugging and depth camera modes allowed us to optimise every aspect of the interaction experience. Our development team provided regular “work in progress” videos and executable versions for the client allowing feedback to be incorporated at every stage.

Inition were present for repeated on-site testing with JCDecaux prior to deployment to ensure the software could cope with the rigours of the real-world environment. Issues included reflection scatter from glass shop windows which were tackled with temporal and contrast filtering of the depth image.

Check out the video above to see the installation in action!

When the producers of North One Television approached 3D creative services company Inition to provide the technical expertise for an ambitious 'virtual presenter' challenge, LILA was an obvious solution.

LILA is a portable motion capture solution which was used to capture the motion of The Gadget Show presenter Suzi Perry in London and transmit that captured data to drive an avatar of Suzi who was composited in The Gadget Show studios, all in real time.

Marker-less Motion Capture

Based on real-time 3D animation software Autodesk MotionBuilder, LILA is Inition’s marker-less motion capture solution which can provide show stopping interactive installations for brands, exhibitions, tours and marketing agencies.

Without any calibration required, Suzi stepped into the holo-deck style cube at Inition’s London studios where her silhouette was tracked by 14 x 2D cameras. No markers or special clothing were required. The vision processor software calculated a 3D mesh of her body from these views of which a digital skeleton was applied using the Organic Motion plugin.

The frame-by-frame motion co-ordinates from the data was then transmitted to the MotionBuilder system at North One Television studios in Birmingham via a special driver which then applied Suzi Perry’s features such as clothes, hair, skin tones etc in real time. The avatar of Suzi which was composited into the studios of The Gadget Show next to the real co-presenter Otis and an android version of Jason Bradbury who she was competing with.

Suzi’s avatar character included full face expressions and mouth shapes driven by Suzi's voice fed through LILA's voice recognition software.

The composited footage was sent back to Inition’s London studios via Skype which Suzi could monitor on a screen seeing exactly what the public would see.

A World First

This project posed several challenges, not least because of the processing power required to deliver the avatar in real time. This was the first time a motion capture system had been used to drive another avatar over the Internet.

The LILA motion capture is a complete attention grabbing solution which is available for events, promotions and exhibitions. As well as hire of the LILA system itself, Inition also provide a complete production service with the hardware, software, and creative design of the characters, environments and interactions.