Perhaps this is just me, but reading

my reaction is

Why if x0 and x1 are zero? If we're encoding an image with zero width, then use the full image size? That seems odd. Also, what if it is zero height?

The code has been like this since PIL. I suspect the intention was to handle

where no extents are passed in, and say that in that scenario, the extents should default to the entire image.

I think it would simpler if we just explicitly passed that in from Python.