Converting device independent pixel to device dependent pixel?

Visualizing 100.000 xy-data points with increasing x values
by using a Polyline takes a noticeable amount of time - espc.
if you have to visualize several data series in one chart control.
Any way it's a foolish thing to do so, since the area I am using
to plot the data might only be 1000 pixel wide.

In the old days of GDI(+) I therefore reduced the number
of data points by evaluating which points will be transformed
to the same pixel column.

But now I do not know anything about the real pixels at all!
OK, I could make an estimation and reduce the data as if
there were let's say 1000 pixels and this works fine, but isn't
there a way to exactly convert a device independent pixel from
a canvas to an absolute pixel position on the screen?

Thanks,
fö

I don't think there is a way to use device dependent pixels.

What you can do is to use the fact that a DIP (device independent pixel) is
1/96 inch so any two points that are too close (for example 1/10th of a pixel
- that means 1/960 inch) will be rendered as one.
--
Valentin Iliescu [MVP C#]


"fö" wrote:

> Visualizing 100.000 xy-data points with increasing x values
> by using a Polyline takes a noticeable amount of time - espc.
> if you have to visualize several data series in one chart control.
> Any way it's a foolish thing to do so, since the area I am using
> to plot the data might only be 1000 pixel wide.
>
> In the old days of GDI(+) I therefore reduced the number
> of data points by evaluating which points will be transformed
> to the same pixel column.
>
> But now I do not know anything about the real pixels at all!
> OK, I could make an estimation and reduce the data as if
> there were let's say 1000 pixels and this works fine, but isn't
> there a way to exactly convert a device independent pixel from
> a canvas to an absolute pixel position on the screen?
>
> Thanks,
> fö

To decide what "too close" exactly means I would need to find out
the current screen resolution and DPI settings. Is there a way to
get these values?

"viliescu" wrote:

> I don't think there is a way to use device dependent pixels.
>
> What you can do is to use the fact that a DIP (device independent pixel) is
> 1/96 inch so any two points that are too close (for example 1/10th of a pixel
> - that means 1/960 inch) will be rendered as one.
> --
> Valentin Iliescu [MVP C#]
>
>
> "fö" wrote:
>
> > Visualizing 100.000 xy-data points with increasing x values
> > by using a Polyline takes a noticeable amount of time - espc.
> > if you have to visualize several data series in one chart control.
> > Any way it's a foolish thing to do so, since the area I am using
> > to plot the data might only be 1000 pixel wide.
> >
> > In the old days of GDI(+) I therefore reduced the number
> > of data points by evaluating which points will be transformed
> > to the same pixel column.
> >
> > But now I do not know anything about the real pixels at all!
> > OK, I could make an estimation and reduce the data as if
> > there were let's say 1000 pixels and this works fine, but isn't
> > there a way to exactly convert a device independent pixel from
> > a canvas to an absolute pixel position on the screen?
> >
> > Thanks,
> > fö

Finally I found the answer myself: the classes PresentationSource
and CompositionTarget help to get a transformation from DIPs
to pixel and vice versa. Here's a function that returns the resolution
of a canvas' target device in dots per inch:

public Point GetResolution(Canvas canvas)
{
Point dpi = new Point(96, 96);

PresentationSource source = PresentationSource.FromVisual(canvas);
if (source == null)
return dpi;

CompositionTarget target = source.CompositionTarget;

Matrix m = target.TransformToDevice;
MatrixTransform t = new MatrixTransform(m);

Point pt1 = new Point(0, 0);
pt1 = t.Transform(pt1);

Point pt2 = new Point(96, 96);
pt2 = t.Transform(pt2);

dpi.X = pt2.X - pt1.X;
dpi.Y = pt2.Y - pt1.Y;
return dpi;
}


"fö" wrote:

> To decide what "too close" exactly means I would need to find out
> the current screen resolution and DPI settings. Is there a way to
> get these values?
>
> "viliescu" wrote:
>
> > I don't think there is a way to use device dependent pixels.
> >
> > What you can do is to use the fact that a DIP (device independent pixel) is
> > 1/96 inch so any two points that are too close (for example 1/10th of a pixel
> > - that means 1/960 inch) will be rendered as one.
> > --
> > Valentin Iliescu [MVP C#]
> >
> >
> > "fö" wrote:
> >
> > > Visualizing 100.000 xy-data points with increasing x values
> > > by using a Polyline takes a noticeable amount of time - espc.
> > > if you have to visualize several data series in one chart control.
> > > Any way it's a foolish thing to do so, since the area I am using
> > > to plot the data might only be 1000 pixel wide.
> > >
> > > In the old days of GDI(+) I therefore reduced the number
> > > of data points by evaluating which points will be transformed
> > > to the same pixel column.
> > >
> > > But now I do not know anything about the real pixels at all!
> > > OK, I could make an estimation and reduce the data as if
> > > there were let's say 1000 pixels and this works fine, but isn't
> > > there a way to exactly convert a device independent pixel from
> > > a canvas to an absolute pixel position on the screen?
> > >
> > > Thanks,
> > > fö