I’ve spent the last week madly trying to come to grips with coding in C and then learn C++, in an effort to get down and dirty with OpenGL. Doesn’t make for a particularly interesting journal entry. Fortunately though, my new band-pass filter arrived today, for the soon to be cracked open PS3 Eye.
To the dismay of my parents, I pulled apart just about every toy I owned as a child. Just to see how they worked. Rarely could I put them back together, but that was beside the point. Thanks to an excellent video tutorial, this little expedition went (mostly) according to plan.
Aside from the joy of breaking toys, the reason for pulling apart the PS3 Eye and replacing the lens is so that I can track movement in the dark more easily. Obviously light projections aren’t very effective during the day, so my work will be shown in low-light conditions. Normally, you’d want a camera to see what the human eye sees – what we consider ‘visible light’. Infrared is outside of this spectrum, but isn’t outside the vision capabilities of a digital camera.
The way a camera like the PS3 Eye deals with this is to add an infrared filter, so that it can only see visible light. It was my duty to seek and destroy this filter and replace it with one which did the exact opposite: block out all other light in favour of infrared frequencies.
I certainly wasn’t as calm and collected about the whole process as the guy in the video, but I did manage to bust open the casing, only inflicting a few wounds to its black plastic body. What was really (really) difficult, was getting the infrared filter out of the camera. For anyone thinking about doing this for themselves – please wear glasses. I’m being absolutely serious about this and it’s probably the only real omission from the video tutorial: the lens is glued in quite well and you’ll find tiny pieces of plastic, glass and metal blade shooting toward you as you try and pick the little bugger out…
Finally, I replaced the tiny infrared filter with a much, much larger band-pass filter. I’m a little concerned that I may not have done this part quite right, as it’s now sitting directly on top of the camera sensor, making me a little concerned for its safety. It did all fit back together though, and I guess I can’t really roll with the DIY crew until I break a few things anyway.
I’m yet to check improvement for motion tracking, but a quick look in MaxForLive (with the help of macam) shows that the filter has done its work. I now look like I belong in a cheap horror film, jumping around wildly in black and white…
This video is simply to display the different amounts of IR light that come from different sources. Clearly daylight and my desk lamp have plenty, but interestingly there is nothing coming from the computer screen (yes, it’s switched on). This is a really helpful property, because it removes any moving light sources that may distract the motion tracking software – in my case, light projections.
It won’t be until I build an infrared floodlight next semester that this camera will really come into its own, but I do plan on doing some small scale testing over the next couple of weeks – just as long as C++/OpenGL doesn’t steal my sanity first.