Mixing matplotlib transforms

Question

I would like to add a patch, such as Rectangle, to a Matplotlib plot where the position of the patch uses ax.transData but the size is in pixels or points. Is there a way to set up a transform that works that way? Or some other way other than transforms?

Answer 1

This answer probably comes a little too late, but if someone else stumbles upon this question:

There is an example for this in the Matplotlib documentation:

import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import matplotlib.transforms as transforms
fig, ax = plt.subplots()
xdata, ydata = (0.2, 0.7), (0.5, 0.5)
ax.plot(xdata, ydata, "o")
ax.set_xlim((0, 1))

trans = (fig.dpi_scale_trans +
         transforms.ScaledTranslation(xdata[0], ydata[0], ax.transData))

# plot an ellipse around the point that is 150 x 130 points in diameter...
circle = mpatches.Ellipse((0, 0), 150/72, 130/72, angle=40,
                          fill=None, transform=trans)
ax.add_patch(circle)
plt.show()

So basically you encode the position of the patch in the transformation and then draw the patch at position (0, 0).