Direct all to imshow fails when calling with a byte format image

[cwgamma]

I create an image, mapping data in the range 0->1.0 through the colormap

import matplotlib as mpl
cmap = mpl.cm.get_cmap('jet')
img_rgba = cmap(dt, bytes=True)
The shape of the array is (2687,2456,4), then I scale the RGB values so that the array is then reduced to (2687,2456,3)
then:
fig = plt.figure(figsize(fw,fh)
ax1 - fig.add_axes([0.0,0.0,1. 0.95])
cax modest_image.imshow(ax1, img_rgba, interpolation=imode, vmin=0, vmax=255, cmap=cmap)

but at the end, when I call plt.show()
I get this error:

/Users/cw/anaconda3/lib/python3.5/site-packages/matplotlib/image.py:328: RuntimeWarning: divide by zero encountered in double_scalars
in_bbox.height / A.shape[0])
Floating point exception: 8

What could be the problem?

[cwgamma]

The previous version of modest image, worked in my progam however nearest-neighbor interpolation (subsampling) was not working correctly while the matplotlib imshow was working correctly, just excruciatingly slow. The test program in the new modest_image package appears to work correctly. I tried using the updated modest_image on both python2 and 3.