Has this edge detection just gone hyperbola?

[vnp85]

I'm trying to "fool" JSolEx into processing even the pixel shifts that contain "only a partial disk", because the parts in question contain a nice doppler. To attain this, I vertically padded the .ser video in pipp. However, I got this edge detection.... and the pixel shifts close to zero are missing there... equator.

[vnp85]

the above is an edge detection for "zero" hydrogen alpha, and this one is at a pixel shift (note: the data is not mine this time, I am trying to process it)