• Matplotlib 1.5.1, Python 2.7.11, OSX 10.11.3
• installed via pip, conda virtual env

Is the rendering of a map (or collection of polygons) in MatPlotLib hardcoded? I'd like to parse out the polygons for interaction, and was exploring ways do it with matplotlib and mpld3 extensions. Parsing out points in a scatter plot is no problem, but the polygons only show lat/long coordinates, where I'd like to hover on a specific polygon and pull its attributes and info. From mpl3d documentation there were some comments of hardcoding certain parameters, so perhaps that's the same issue? In other words, I'd like my vector graphics appear as vectors, not flattened rasters.

I also tried adding a line to force rendering as an SVG image, to potentially parse the polygons that way, but had the same result. (%config InlineBackend.figure_format = 'svg' ) It looks like a file i/o workflow would be required to get get around this, but ideally could visualize in matplotlib all the way (or at least in a way that could be easily pulled into mpld3).