I'm doing some work with an accelerometer - using an iPhone as a granular sampler, basically - and I'm trying to make a 'zeroer' device, so that I can choose what position of the phone means 'play the sample normally'. My problem seems to stem from the way the accelerometer outputs data - for instance, moving the phone in a circle around one axis gives as x co-ordinates 0 to 1 in the first quadrant, 1 to 0 in the second quadrant, 0 to -1 in the the third quadrant and -1 to 0 in the fourth quadrant. I basically just want to add, say, 0.2 to all the data, but have it stay with in the range and change direction when it's supposed to....

...This has been causing me far more problems than I feel it should... any suggestions?

Thanks

Adam

Try using autoscale,

I'm not sure wether it's in pd extended so I attached a zip of the abstract and the help file.

http://www.pdpatchrepo.info/hurleur/autoscale-1.6.zip

Thanks - I'm actually using vanilla so I can run my patch on an iPhone via RjDj, which rules out the lovely expr object in that autoscaler- however, Andy Farnell has made a lovely autoscale object so all's well...

Unfortunately I can't figure out how to solve this problem via autoscale. The data from the accelerometer for each axis cycles up and down between -1 and 1, in such a way that two different positions on the x axis actually give the same value, and it's only by looking at the other axis or something that you can figure out where the phone is...

I think I'm trying to apply a phase shift to the data. But I'm not sure...

If anyone has any experience with processing / dealing with acceleroemter data please let me know! I can't think in 3 dimensions...

Thanks

Adam

Hi Adam,
the [m_scale] object from the rjdj library should do what you want. It "scales input range to output range with linear or exponential interpolation" from high to low and the contrary, so you can do everything you want.
About the iphone accellerometer i'm not so shure it gives values from -1 to 1.
In many examples of the RjDjComposerPack you can find values from -2 to 2 and i saw -3 to 3 somewhere but i don't know how to find out the real ithing output values.
I'm interested in the rjdj tools as well so let me know.

Hey Adam,

Say we are looking solely at data in the x axis. As you pointed out the data moves from 0 -> 1 -> 0 -> -1 -> 0. Which would look like a sine wave if plotted on a graph.

So there are two positions when tilting the device that output the same x value. In order to work out the orientation you'll have to combine that value with another axis value. For example the z axis will show 1 or -1 whether the device is facedown or face-up (can't quite remember if that is the correct order).

However, there are some positions where the accelerometer information is not adequate enough to calculate. If the device is flat on a table it would not be possible to see if the device is rotating. In this case the compass data (3Gs+) has to be used.

Additionally the gyroscope (4 only) can be used to determine rotation in the x axis when the device is vertically up or down.