Linked below are some early results from a new series of sketches I've been working on using Processing. These sketches continue in a long line of projects I've completed recently using simple camera tracking algorithms to infer interesting patterns of movement in urban spaces.
The first example is a calibrated blob tracking experiment, using the excellent and very well documented OpenCV library for processing. A few simple modifications to the setup parameters allow for a very customisable tool, able to withstand many of the constraints live webcam installs can throw up. I've tested this in a number of places (my bedroom wall, lit by a single lamp tends to be the best contrast) and will have more to say on the nature of live webcam video in the future.
OpenCV blob tracking - calibrated from Jason McDermott on Vimeo.
The second example is a first attempt at combining the live blob tracking with the wonderfully funky and playful MSA Fluid library also for processing. This lib is geared towards touch screen interfaces and screen based mouse interactivity - but I immediately thought it would be the perfect partner for my webcam based projects (or even accelerometer/phidget/slider/midi sensor data). It wasn't very difficult to swap out the mousex/pmousex variables for centroid x/y data, so the first test has been deemed a success. I showed this yesterday to Frank/Ale/Amy/george/anyone who would stop for more than 2 minutes in the interactivation studio and it was a big hit 
OpenCV + MSA Fluid (Processing) from Jason McDermott on Vimeo.
The third example is significant for a couple of reasons - it is another combination this time using recorded video of an actual installation space (filtration fields / DAB courtyard) thus requiring another version of the calibration - but also my first experiments in putting together an arrayed interface between the blobs and the fluid.
To explain further; Firstly it's easy to switch out the mouse for 'something else' and inferring movement velocity for a single object/blob is simple. Secondly I wasn't so sure about the way to apply this singular blob mousex/pmousex-esque technique to many objects at once. Thirdly I wasn't sure if it would all explode in one big fluorescent, particle mess!
OpenCV + MSA Fluid (processing) test 3 from Jason McDermott on Vimeo.
[Update]< <note, v3 will be embedded when vimeo uploads my video. since when does a new video have to wait in a queue for 30 minutes??>>
In the end I'd say it's mission accomplished, certainly with calibration tweaks to occur before I'm happy to unleash this on an unsuspecting public. I'd be interested to see how this could influence peoples' behaviour in the space - whether or not we would see people dancing/swimming/painting the space of the courtyard. I'm curious also to see how this kind of new interaction with the space of the DAB could filter into a new perception of the building as not merely a space to move through but one which is open to new forms of physical conversation.
Related Posts
architecture com dab data float gallery google information interface j line pixeltag processing 39
Comments
2 Comments so far. Leave a comment below.Hi Jason
I am currently working on an installation which controls synthesis parameters using motion tracking in Jitter and who like to add a visual aspect to the piece.
I have only recently become aware of Processing and am new to programming language. I have had many failed attempts to replicate your efforts on combining Open CV and MSA Fluid.
Would you be willing to share your Processing sketch for this? Any and all reference to this element of our final installation would of course be credited to you
Thank you
It is useful to try everything in practice anyway and I like that here it’s always possible to find something new.