Last week I rebuilt my air quality sensor with the fan fitted right on the side of it. I’m now getting results that seem qualitatively in line with expectations plus I got to play with multimorph plastic and a hobby drill so that’s good. I re-wrote a fair chunk of my kinect code and got it working more stably (but black jumpers are still a problem) but I’ve decided to shelve that for the time being and spend some more time concentrating on recording some sound and making it work musically with various inputs…
Last week I mostly tried to perfect my breathing detector in openframeworks. The conclusion I came to is that it works really great if you’re wearing the right coloured jumper and not at all if you’re wearing black or even worse a checked shirt. Having booked myself in for a Skype chat with the ever helpful and talented Zach Lieberman I think the dark clothing can probably be overcome by improving the contrast and standardising the input somewhat. I don’t think the checked shirt problem will be overcome but you can’t please all the people all the time. Given the installation will do other stuff in response to movement and standing still besides responding to people breathing I’m not too worried. Besides, check shirts are all ghastly.
I also boxed up my home made air quality sensor and tried to calibrate it with an expensive one kindly loaned to me by the University of Birmingham. I left it running all day and burned some incense next to it to give it some nice tiny lung ache inducing particulates to measure. Sadly when I compared the results I found my sensor was detecting something very different, so different in fact it made me question the validity of the results at all. Having read through the literature again I’ve decided I am probably better off sticking to my original plan and trying to recreate the work of indiairquality blog by drilling a hole in the side of the PPD42 and fitting the fan to suck air through it, alas the laminar flow light locked laser cut case might have been a blind alley.
I’m at that irritating middle part of any creative endeavour where it seems like most things are going wrong and you start to question the validity of your original idea and your own ability as a creative practitioner. This is fairly standard though so I’ll just muddle through. I went to a good artist’s talk by Nicky Pugh recently where she mentioned the importance of play testing ideas early on and regularly through out a project so I shall get on with that and try to find ways of solving my kinect and air quality sensor problems while play testing my ambient music generator with the kinect…
Last week got lost entirely to laser cutters, making a case for my sensors and porting the code for my measuring breathing trick to the kinect.
On Wednesday I visited FizzPop the Birmingham maker space to monopolise their laser cutter for a bit and curiously bumped into two guys also building home brew air quality sensor networks. They were both using laser particle counters that were slightly more expensive than mine. I got some really good tips on how to get the best from my sensor by placing it in a light locked box with rounded corners for the air flowing through it to stop any particulates getting dumped on the corners as they pass in.
Apparently preserving a laminar air flow is important in getting the best from these cheap sensors. In the end I was convinced by the argument enough to ditch my original box idea, modelled on the cases I’d seen at indiaairquality blog and try something along those lines.
The rest of the week I had my head stuck in openframeworks with the kinect. I’ve got something that can reliably tell when someone’s breathing when they’re close to the kinect’s camera and I’m currently fine tuning my algorithm so it will work better at a distance.
Away from BOM lab at home I finished (mostly – I still want to edit it down) another piece of music based around a field recording, it’s not as ambient as I see the material I want to write for the sound installation but these things often take on a life of their own. It was inspired by the bird song in the field next to my house but it ended up sounding like some clangers jamming with rubber bands and a cat. I’m thinking more ambient less time, need to turn off the drum machine….
A third of the way in! By Friday I’d built a portable air quality sensor that works with a lithium ion battery and an SD card logger but no sign of a nice laser cut case. I’ve had to make do with my lunchbox with a few holes drilled in the lid for the time being. This week I’ll hopefully get to calibrate it with the help of environmental scientists from Birmingham University. Full credit to the lad (he doesn’t leave a name on his site) who wrote up how to adapt the sensor I’m using for greater sensitivity.
When I wasn’t spending time soldering and desoldering things I coded an openFrameworks program that does a pretty good job of identifying when someone takes a breath from video footage from the web cam alone. It works well if people stand still but is totally useless if they’re moving but I’ve designed my interaction scheme around this so I’m confident it’ll still be useful. I’m spending this week setting it up to work with a kinect, which has thus far been a total nightmare.
I also did some more field recording and tried to go full ambient in the studio over the weekend. Unfortunately I picked up my drum machine and 303 and ended up making really weird microtonal xenharmonic acid instead so I’ll have to go back to the drawing board there….