Having trouble with a standard LDDMM registration, example provided

[ghost]

I am running the following script to perform a simple LDDMM registration between two images:

I0_obj = Image.open('I0.tif').convert('L')
I0 = np.array(I0_obj)

I1_obj = Image.open('I1.tif').convert('L')
I1 = np.array(I1_obj) 

transform = ardent.Transform()
transform.register(I0, I1, track_progress_every_n=10)

forward_warped = transform.transform_image(subject=I0, deform_to='target')
forward_warped = Image.fromarray(np.uint8(forward_warped))
forward_warped.show()

This displays the following warped template as output, which does not appear to be anywhere near my desired I1 target.

For reference, here are my original I0 and I1 (template and target, respectively):

From the images above, it's clear that some deformation is taking place, but it does not seem to complete. What might be the issue here? I've attempted using a larger num_iterations and increasing the sigma_regularization, neither of which had any effect on the output.