Hi,
I'm looking at gnuplot svg output with a high zoom ratio that makes it
easy to examine accuracy. There are discrepancies.
http://i47.tinypic.com/1zcpjtc.png
this screen shot shows such a zoomed area and using mouse cursor feature
to find the y coord of the grey line I get a reading of 0.49
In this case the textual dot that marks the current cursor readout is
squarely on top of the grey line and of the same size as the line width.
The next quantised cursor position puts the dot just under the grey line
up against it. It shows a y value of 0.21; when above and touching is 0.76
As far as I can tell by eye it is the lower edge of the square textual
dot that represents the value shown in the mouse cursor read out. This
also seems close to the tip of the arrow shaped mouse cursor in Firefox
that I am using to render the SVG. I get very similar results on Opera.
This raises the question of whether the mouse readout is an accurate
value for the position of the grey line.
Is the plotted line draw equidistant from the data value (ie data value
is at the centre of the draw line - as one would probably assume) or is
the data value at the lower extremity of the plotted line?
In this example the line would seem to be closer to 0.75 but cursor read
out is 0.49
If I try to get a readout on the x axis, the closest I get to zero is
-0.03 when the bottom of the mouse cursor is near the centre of the line
drawn as the x axis.
Taken together, this would suggest that the plot line is drawn on and
above the data value , not centred on it. These lines are default line
width.
The result is that the plotted line appears to higher than the actual
data value and placing the cursor at a position on the graph gives a
coord readout larger than the data value unless on realises that it is
the lower edge of the marker dot that should be used to indicate the
position.
The "dot" referred to here is the first character of cursor readout
text, presumably this is a period (or full-stop) character.
Regards, Peter.
|
