adrian cockcroft adrian.cockcroft at
Tue Apr 3 19:32:23 CEST 2007

Two more thoughts...

If you hold down one finger and tap the other one, the mouse pops over
and back again. This could be a way to set a bounding box or turn on
the mode.

We may be able to set the sensitivity and sampling rate much higher
than normal to support a more advanced algorithm.

To test this, we could capture and share some raw (time,x,y) event
streams from a touch screen, and try processing them offline. Then we
don't need actual hardware (a spreadsheet is may even be good enough)
to figure out algorithms.


On 4/3/07, mathew davis <someoneinjapan at> wrote:
> On 4/3/07, adrian cockcroft <adrian.cockcroft at> wrote:
> > This is key:- "Pressure has almost no effect on a single touch, but
> > not so on a double touch. The relative pressures will cause a
> > significant skewing effect towards the harder touch. You can easily
> > move the pointer along the line between your two fingers by changing
> > the relative pressure."
> > So we will not see clearly defined bounding box limits. The point will
> > skate around within the limits depending on relative pressure.
> So I guess we need someone with a device to test this and see how much
> pressure actually affects the neo touch screen.  With that information I
> think we could see how easy it would be to get an average bounding box
> approximation..
> > The first finger will set a clear start point, the second finger will
> > make that point shoot off towards it, but it will not go all the way
> > to the second touch. The effect should be to oscillate along the line
> > between the two end points, and it wont return to the position of the
> > first touch.
> >
> > If we capture a clear single touch, and an average position of
> > oscillation, then we can take the average oscillation to be the center
> > of the bounding box, and project an estimate of the opposite corner
> > where the second touch should be. With the right filtering and
> > limiting algorithm it should be possible to get the effect we want. If
> > we can give visual feedback on the screen showing the touch points and
> > bounding box it may help the user control the input better.
> Ok so what if this feature was disabled by default.  Since enabling it might
> slow some functionality.  When the user enables the feature he/she will have
> to go through a config which does a calibration.  The user runs through
> several scenarios where the program can gather the relative pressure
> difference in known circumstance with the desired result known as well.  It
> could then store that information in a database based on which user is using
> it, if there are multi users, if not just store it in a config file.
> > Challenges: In comparison with a true dual touch input device, its
> > going to react more slowly as the algorithm will need to gather more
> > data to decide where the pointer should be. Some of the faster moving
> > single touch gestures may be hard to distinguish from multi touch.
> >
> > Adrian
> I agree, but it would still be a nice feature to have and I could deal with
> a little lag for added feature especially if I could enable it or disable
> it.  With the single touch fast gestures we could set that up inside of the
> calibration also.  That way if it thinks it's multi touch it compares it
> with it's database which links it to the single touch fast gesture instead.
> It would be slow but you could disable the multi touch option and then
> single gestures would be fast again.  Or we could have the multi touch be
> enabled in certain programs like web browsing or picture viewing, and
> disabled for the rest.  But I think a simple button in the header or footer
> should work alright.  You could have it turn green when active and red when
> deactivated.

More information about the community mailing list