Acknowledging a sudden increase

    Joe
    Jun 21 2013 | 4:24 pm
    Hi
    This is a difficult to explain but basically I have a numerical output which is always changing. However, I dont want to acknowledge the small changes in numbers, but only when the numbers increase quickly and a large increase of numbers.
    If anyone has any ideas if this is possible, id appreciate the time,
    J
    • Roman Thilenius
      Jun 21 2013 | 4:36 pm
      it really depends on if we are talking about consecutive numbers, about a certain time period, or about a tendency, the latter which would require that the user may (and have to) set the time and difference values arbitrarily.
      for consecutive numbers, something like [if abs($f1-$f2)>=50 then 1 else 0], where $f2 is the last value, will be all you need.
      if time is involved, [timer], and probably [accum] and [mean] should be created in your patch now as a precaution. ;)
      -110
    • Joe
      Jun 21 2013 | 8:31 pm
      I don't really know which of those options my situation fits into sorry.
      If this helps, the numbers are the y corridents produced by blob tracking. So as you can imagine the numbers are constantly changing...
      However, what I want to do is create a way of making the patch understand when those numbers rise or fall is over a particular range at a particular speed
      I can't quite get my head round that at the minute. Does it need to be something like, if a change over e.g 10 happens within 50ms then bang? does that make sense?
    • Roman Thilenius
      Jun 21 2013 | 11:17 pm
      "Does it need to be something like, if a change over e.g 10 happens within 50ms then bang? does that make sense?"
      yes, of course. just calculate a "velocity" value from these time and difference values, then work on that.
    • brendan mccloskey
      Jun 22 2013 | 11:41 am
      Hi here's a Max solution (as opposed to a Maths solution); using [trigger], [pipe] and [-] to give you velocity. You will need to tweak a few parameters, as Roman states, to get the desired result.
      There is a related thread here:
      https://cycling74.com/forums/simple-problem-detect-peaks-troughs-in-sine-wave-in-max-not-msp/
      Brendan
    • Joe
      Jun 22 2013 | 12:01 pm
      Wow, that is brilliant. Needs a bit of tweaking to fit my patch but that is spot on. Thanks
      J
    • brendan mccloskey
      Jun 22 2013 | 12:17 pm
      :) this forum is brilliant