Dear fellow ChucKists,<br><br>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.<br>
<br>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.<br>
<br>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.<br>
<br>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.<br>
<br>Hope that was of any use/interest.<br>Yours,<br>Kas.<br>