Re: [Matplotlib-users] Real time graphs

[Mark S. <ms...@bi...>]

Hi,

That seems to have helped a fair bit, thanks for the tips.

Mark

On Fri, 2005-06-03 at 22:34 -0500, John Hunter wrote:
> >>>>> "Mark" == Mark Saward <ms...@bi...> writes:
> 
>     Mark> Hi, It looks good thanks, but I'm having a couple of
>     Mark> problems (being a python and matplotlib newbie).
> 
> You can use the "GUI neutral" pylab interface to do animation, but I
> don't recommend it.  This is mainly for teaching, demonstration and
> testing purposes.  For real work, I would settle on a GUI toolkit and
> use their event loop (idle handling or timer) to do the animation.  If
> you are able to choose, I recommend GTKAgg as a backend for animation
> because it is fastest (fltk is close, I haven't profiled qtagg; wxagg
> and tkagg are significantly slower).
> 
> See, for example, examples/dynamic_image_gtkagg.py for an example of
> how to use the gtk idle handler to do animation.
> 
>   Mark> 2.  It flashes between the specified axis and just fitting the
>   Mark> graph to the boundaries defined by the actual data.  I want it
>   Mark> to stay using the specified axis all the time
> 
> I suspect this is because you are not properly controlling when
> drawing is done.  For example, if interactive mode is on and you do
> 
> for frame in myloop:
>   axis(*rect)
>   plot(blah)
> 
> your figure will be drawn twice.  Once when axis is called and once
> when plot is called.  plot calls autoscale and may change the axis.
> What you want is to turn interaction off and explicitly control the
> time of the drawing with a call to canvas.draw() or pylab.draw()
> When interaction is on, all pylab commands (including axis and draw in
> this example) trigger a call to draw,
> which is not you want.  What you want is something like
> 
> # turn interaction off
> for frame in myloop:
>    plot(something)  # or set your line data, whatever....
>    axis(*rect)
>    draw()
> 
> Here you only draw after the axis limits have been set.
> 
> Actually, in animation you rarely want to call plot for every frame.
> It is much more efficient to save an object and set its data, as in
> dynamic_image_gtkagg and anim.py.  Calling plot in every frame creates
> a new Line2D object on each iteration of the loop and is slow.
> 
> Hope this helps,
> JDH