2.33D Sketching

Why did I do this?

Absolut tracking in realtime is still a big challenge. Cameras with over 100 fps are mostly for industrial or military use and are cost intensive. However there are some alternatives. E.g. The PS3 Eytoy 3 Camera can deliver up to 60 fps for a very competetive price and Nintendos WiiMote IR-Cam delivers even 100 fps. In general I am especially interested in tracking ranges between 8 - 60m for outdoor use with an accuracy of approximately 10cm. Even so we are not even close to that in this project, pairing up the WiiMote IR-Cam with Gadgeteer gives us a very good platform for experimenting around with the camera parameters. In prior testings I had tracking results up to 25m with a custom active IR target.

How was it done?

Opening the WiiMote is a rather difficult task, because Nintendo still uses its custom screws to protect them from easy opening. However, I bought recently a screwdriver for their screws to get over that problem. They are very rare and if you can't get one, use a standard small screwdriver and force it in. The screwdriver will bend a bit, especially if u use low quality ones. For extracting the WiiMote camera put a large solder blob on all pins so that you can heat them up at the same time. Pull gently and you are done. Connecting the Cam is really straigth forward. It is a 3V3 cam that needs pullups on SDA and SCL. The oscillator I used is a 24MHz, 4 Pin one. It needs 5V on the input pin to deliver a good signal on its output for the cam's CLK. The Reset Pin of the cam has to be pulled up to activate the cam.

A first demo

For first testings I dug up my old actuated IR-Pen I built some years ago. Amazingly it still worked. The sample program I wrote is fairly easy. The WiiCam provides 3bytes of information for each of four possible IR-Blobs. 10bit for x, 10bit for y and the remaining 4bit for the Blob size. The Blob size is used in the demo to change the brush size of the pen.

Result

It is amazing to see how fast the system reacts to the hand movements. In the demo the speed was limited to sampling frequency of 100 Hz and no camera external filters or estimation algorithms were used. The data output of the cam is really amazing and one has not really to care about any noise problems. I will experiment with some other the camera initializations in future, to extend the maximal tracking range.