Re: [Matplotlib-users] How to blank an area of the canvas?

[Brendan B. <bre...@br...>]

Brendan Barnwell wrote:
> 	I'm trying to find the quickest way to erase a rectangular area of 
> the figure canvas.  I tried using canvas.restore_region with the 
> optional bbox argument, but there seems to be some mismatch between 
> the measurement units of the saved buffer object and the currently 
> shown data.  For instance, if I have a Text object on my plot, I tried 
> this:
> 
> bbox = g.text.get_window_extent()
> canvas.restore_region(background, bbox)
> 
> . . . but it does not correctly block out the text.  (The restored 
> rectangle from the background appears elsewhere on the axes.)  How can 
> I convert the buffer coordinates to the coordinates of the the 
> displayed plot?

	I'm sorry to bump my own post, but I would really appreciate some 
help with this.  I've been wrestling with it for a couple days now, 
and I cannot figure out how the coordinate system of the saved canvas 
is related to the axes coordinates.  I have found that with 
bbox.transformed(ax.transData) I can at least get the coordinates 
scaled to fit on the axes, but they are still offset in position from 
where the box actually appears on the canvas.  I can't figure out how 
to compute this offset.

	By playing around with the coordinates manually, for instance, I've 
found that adjusting x by -52 and y by 21 appears to line up the 
canvas with the axes, but I can't see where these numbers -52 and 21 
would come from.  My saved canvas buffer's get_extents() method 
returns (65, 50, 586, 443), so I thought that the appropriate offsets 
would be 65 and 50, but that doesn't work.

	So, what coordinates (x1, y1, x2, y2) do I need to use in 
canvas.restore_region(savedBuffer, (x1, y1, x2, y2)) in order to 
restore precisely the area of canvas occupied by a patch drawn at axis 
coordinates (a1, b1, a2, b2)?

Thanks?
-- 
Brendan Barnwell
"Do not follow where the path may lead.  Go, instead, where there is 
no path, and leave a trail."
    --author unknown