WiiLi.org

[blahblah]

Surely you can get the roll with just two points? If you fire up wiinremote, it gives you a visual of the ir points and they roll if you roll the wiimote. I'm away from my wimote, but this is what I recall. Presumably you only lose this if you have the camera directly on the axis of the two sensors.

[CarlKenner]

With 4 IR points you can track all 6 DOF using IR, IF you can tell which dot is which.

Actually the P5 Glove uses 3 IR dots for 6DOF, but the P5 glove uses a 3D "camera" which also reports depth for each dot.

People's idea of using the IR to seperate rotation from acceleration is excellent. I will try it as soon as I build a sensor bar.

[squeakypants]

[Dodger_]

There appears to be some confusion between blahblah and squeakypants. Squeakypants is talking about using the IR data from 3 points to negate any accelerations that are experienced when the Wiimote is rolled. This would give you much more accurate acceleration data. Blahblah is speaking about the much simpler process of getting JUST roll(ignoring acceleration) from 2 data points. Both of you are correct.

[spooL]

ive been thinking about this for a while now, i have been meaning to make a sensor bar but i havent found a place that will sell me ir leds. i think i'll make a bar with some switches that can be a one, two, or three point thing, i think an equalateral triangle would be good.
you can calculate your distance from the bar (or plate i guess) if you know how far they are from eachother, using trig. you can tell if youre viewing them at an angle bu checking how much the triangle is squashed horizontally or vertically.
i wonder if a triangle laying flat on the surface under your screen would work well (im just worried about sticking sensors too high off the top and having them pop out of view).
one ir point is enough to simulate a mouse, if you dont mind the confusion when youre tilted a little.
the triangle sort of breaks down when you go more than 120 degrees, but then any sensor bar wont work when you're not facing it.
i think we could compensate for that with occasional G checks, like whenever the force across all the axes is equal to 1G, record orientation
i'd really like to get a spatial simulation for the wiimote going