The idea was to use the ARTK-Markers and the Augemnted Reality node from vvvv to build a little patch which can be used to produce videos with different faces.
We also used the possibilty to play sounds whenever a specific marker is detected.

Demonstration Video

For the implementation we used the Augmented Reality patch from the book and modified it. We use the Augmented Reality Node to get the position and the ID of the AR-Markers. Then we compare if the markers that we have specified are in the camer picture, and if they are, their position is sent to the 3DWorld Node. In the 3DWorld Node, we get the signal that the marker is detected, we then used two TimeFlop Nodes, an AND and a NOT Node to build a switch which sends only one 1 to the FileStream Node, so that the sound is played a single time and not several times overlapping.
We used PNG images with a transparent background to get faces without anything around it. We put these images as FileTextures on a Box with a really low depth so that we get a really flat plane.
We used the already given GoraudDirectionalSphere Node to get a layer which we can group together with the original Video to get the Image with the exchanged faces.

The sourcecode can be found here:

face-exchanger.zip