Dear fellow ChucKists, This weekend I was experimenting some more with my tilt-sensing joypad, adding a -so far crude- granualar way of playing back samples. When I handed it to some friends to try it turned out people subconciously expected something interesting to happen with vigorous shaking. This sounded like a interesting idea so I made crude and so far badly calibrated implementation. What I'm doing is checking the square of the change in bot axis since the last time I polled them and adding both of those to a "shake" variable from which I'm also subtracting a value periodically (at least as long as it's above 0). The resulting parameter seems to do a decent job of detecting whether the joypad is currently being shaken and if so how hard. As hard shaking tends to build up the parameter I have to let it return to 0 faster if it's higher. The result is -so far- being used to scale randomisation in the grain playback rate. This will need tuning (as well as a cleanup...) and I wonder if left-right shaking may need to be diffent from forward-backwards shaking and how important the period of "re-stabilising" of the whole thing is. So; this is still in a early stage, I only started at it last afternoon but it might be the sort of practical example that works for some people as a illustration of how I try to deal with this sort of subject. BTW, I set this sort of thing loose on my friends because I find observing how they interact with it leads to different insights then I'd get on my own, insights that often make interfaces perform bettter for myself as well. Hope that was of any use/interest. Yours, Kas.