This week I am researching on the possibilities of using motion tracking or object recognition in my project. In short, I want the camera to recognize people and their movement so that their postures will not be affected by special effects. Only the background is going to morph. This technique is possible to achieve in Max/MSP Jitter thanks to 'cv.jit' library. It is all fine except one thing - I do not posses the full version of Jitter which is needed for the library to work. Still, no problem - I can always work in TVU facilities where the full version of Jitter is installed on Mac's. In the meantime, I am researching on similar techniques to cv.jit or an equivalent.
This is what I have found browsing the web. It is called CamSpace and is a great software to play games without controllers such as joysticks, mouse, keyboard or pad. All you need to controll anything in your PC is a camera. This software is recognizing the objects seen by the camera thanks to colour recognition. Anything in front of the camera can become a controller. This is awesome! This invention can be expanded and used in many different ways, not just to play games. But enough reading. See it yourself on their website.It is also ready to be installed on any PC for free.
I found more on this topic of motion tracking. Here is a blog on which the creator is explaining and demonstrating how we can play 'ping pong' live video tracking and a microphone.



I am certain that these inspirations will find place in my project. My idea is to use a statis camera and record a chosen location so to use it as a background of my video. The foreground will be a static image edited frame by frame in photoshop. I am thinking of a city location and using an image of a building in the foreground. Like in the last picture above, I will edit the building 30 times for each second. I am not planing to spend full seven months of editing as Rybczynski did :-) Thanks to mathematical precision of Photoshop effects I can be much quicker in editing.



