convert#

matplotlib_figure_to_numpy#

fury.convert.matplotlib_figure_to_numpy(fig, *, dpi=100, fname=None, flip_up_down=True, transparent=False)[source]#

Convert a Matplotlib figure to a 3D numpy array with RGBA channels.

Parameters:

figobj

A matplotlib figure object

dpiint, optional

Dots per inch

fnamestr, optional

If fname is given then the array will be saved as a png to this position.

flip_up_downbool, optional

The origin is different from matlplotlib default and VTK’s default behaviour (default True).

transparentbool, optional

Make background transparent (default False).

Returns:

arrndarray

a numpy 3D array of RGBA values

Notes

The safest way to read the pixel values from the figure was to save them using savefig as a png and then read again the png. There is a cleaner way found here http://www.icare.univ-lille1.fr/drupal/node/1141 where you can actually use fig.canvas.tostring_argb() to get the values directly without saving to the disk. However, this was not stable across different machines and needed more investigation from what time permitted.