Re: [Matplotlib-users] matplotlib for interactive simulation visualisation

[Chris B. <Chr...@no...>]

AP Meyer wrote:
> My problem is to visualise the process of a simulation, thus vector 
> graphics objects controlled by a simulation program. At the moment I use 
> a Tkinter Canvas, but it is very limiting. Look at the attached image 
> for an example: cars driving on roads and trying to avoid crashing 
> (Windows version). Imagine the cars driving around, of course!

Did you post a similar message to comp.lang.python? Anyway, I tried to 
reply to that, but maybe you didn't see it.

I would suggest my wxPython FloatCanvas package:

http://home.comcast.net/~chrishbarker/FloatCanvas/

(it's also included with recent versions of wxPython)

Sorry I don't have a really nifty page describing it, but it does come 
with a fairly comprehensive demo. I'd describe it's core mission as such:

* It is designed to display data, which are likely to be in arbitrary 
coordinates, rather than a picture, but is not a plotting tool (axes and 
all that)

* It is object based, and thus separates you from most of the need to 
understand drawing with wx (DCs and all), and lets you create and 
maintain objects.

* It is designed to be high performance enough for simple animation

> What I need is something similar to the Java2D API or JGraph. I need 

With a quick glance at the Java2D page, I can say that it has a slightly 
different mission, but the biggest issues are:

- No OpenGL (but I'd love to see that!)

- No printing (though I was given a patch that I have not had a chance 
to integrate yet. It shouldn't be too hard)

- Not as mature or Robust (maybe a half a dozen people are using it for 
real)

As for your requirements:

> - interactive (by application, not user) creation and modification of  
> graphics primitives, such as:
  circles - yes
text -  yes
rects - yes
arcs - no, but it wouldn't be hard to add
arrows - yes, but not in the released version--hopefully I'll get anew 
one out soon, otherwise, I can send it to you directly.

> splines, etc. -- no, but same as arcs.

> - transparency (alpha blending) - No (this is NOT well supported by wx yet, :-( )
> - anti-aliasing -- I think text is covered, but 

These two could be supported by AGG, which is used by matplotlib. There 
was some talk a while back about creating a wxAggDC, which I could then 
use for FloatCanvas, but no one has stepped up to do it yet! Would you 
like to?

> - rotation, skewing, scaling, etc. -- some,a nd it wouldn't be hard to add more.
> - SVG output -- nope

wxArt2d (A C++ library for wx) has a lot of this. It would be great to 
see it wrapped for Python, unfortunately, not a trivial job, as I 
understand it.

Question: do you need it to be user-interactive, i.e. events responding 
to clicking on objects? FloatCanvas does have that.

> I have already looked at Tkinter and wxPython/OGL, but neither seems  to 
> do what I want.

I'm afraid you're going to have to adapt anything that exists for 
Python, the libraries just aren't as comprehensive as to for JAVA, at 
least for this kind of thing. One reason is that there are quite a few 
GUI toolkits, so development is kind of spread thin.

If you really need the alpha blending and anti-aliasing, and matplotlib 
can't do what you want out of the box, I'd highly encourage you to work 
on a general purpose Agg module. It would be a real contribution to the 
community, and I'd integrate it into FloatCanvas, so you could get the 
rest of what you need from that. Another option would be to look at 
wrapping wxArt2d.

-Chris

-- 
Christopher Barker, Ph.D.
Oceanographer
                                     		
NOAA/OR&R/HAZMAT         (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chr...@no...