What you do is you obtain pitch, roll, and yaw from the gravity vector, or low-pass-filtered component, and the X, Y, and Z from the dynamic acceleration, or high-pass-filtered component. It isn't perfect but it does save you a sensor.
I would personally be greatly in favour of a 12db state-variable filter for ChucK with some sort of method to get HP, BP and LP as separate outputs. Maybe following the system set by Pan2 to ChucK multiple outputs to various Ugens?
Do beware of noise from cheap sensors as that will wreck your plan, if my new joypad is any indication.
On the bright side, while exchanging posts with Kassen on the forum, we both separately noticed that there is a trend in the gaming industry, and we are expecting many new 3D mice and joysticks that are based on the now-inexpensive 3-axis accelerometer chips that are available from a few manufacturers.
My personal favourite philosopher Sun Tzu made some commentary about it not being hard to hear thunder. This is generally taken to mean that one should by all means do so but that it's typically a thank-less job :¬)
(I warmly recommend Samuel Griffith's translation and steering clear of the more fashionable commentary's)
And thanks.
p.s. I sent an email to Fender, the famous guitar manufacturer, suggesting the above. Who knows, maybe they'll make a guitar tracker product for us to enjoy.
Weren't they also somehow involved with Guitar Hero? Or was that Gibson? GH, BTW, also uses a tilt-sensor and according to my own research (to use a euphemism) tends to result in a rather active engagement as well. Sadly GH controllers are quite expensive for a few buttons and a tilt-sensor and are a bit small especially if the performer would be -say- 2 meters tall.
Good plans, do keep it up!
Kas.