[Stk] Fwd: Improving support for iOS

[Stephen Sinclair]

On Fri, Sep 13, 2013 at 9:47 AM, Richard Dobson
<richarddobson at blueyonder.co.uk> wrote:
> On 12/09/2013 14:59, Gary Scavone wrote:
>> It would be very difficult to express the controls in physical units.
>> First of all, not all algorithms are physics-based models.  But even
>> the physical models are generally not developed with respect to exact
>> physical measurements and/or quantities.  There is also a trend in
>> the modeling world to use dimensionless control parameters.
>>
>
> Also, a physical quantity is in principle unbounded with respect to
> range and level (e.g. pressure ranging from outer space to bottom of the
> ocean, temperature from absolute zero to the point of atomic fusion and
> above), whereas in any practical system you will be processing (or
> emulating) the output of some physical transducer which will necessarily
> represent some (arbitrarily) reduced and bounded range. In the end, any
> transducer must in principle generate a normalised signal (± 1.0), which
> may or many not be further mapped into a MIDI or some other digital
> range decided by the hardware or software; and thus must be defined and
> configured (calibrated?) by the user.


Uh, sure that's true _in general_, but here we are talking about the
very specific case of musical instruments, where friction
coefficients, pressure, velocity, etc, have pretty well-defined
constraints related to human ability.  Simple selection of appropriate
units could handle any weird cases of too-large numbers.

In any case, as I said in my original email on the subject, I realize
that this idea of physical quantities is not practical everywhere and
would require significant work.  I still think it's an interesting
idea, and could represent a good student class project for instance.
It doesn't have to cover 100% of cases for it to still be useful.  In
my thesis work, I had to do quite a bit of calibration on the "bow
velocity" and "bow pressure" when mapping a haptic force feedback
device to the Bowed object, and I'm still not sure whether what I did
was entirely physically correct.

In principle the normalized/MIDI case is a kind of abstraction over
the physical models that STK implements, designed merely to make
connections with MIDI hardware easier.  Super useful, of course, but I
think exposes the underlying physical quantities, in cases where it's
possible, would be pretty cool.

Steve