In fixing the PSF splitting problem with #10, I managed to create a new problem for the reconstructions (less breaking but it was slicing wrong). It is better to use the header as the ground truth for what the shape should be, so I'm expanding the array out based on the expected shape, and using the header info to determine the channel index.
In fixing the PSF splitting problem with #10, I managed to create a new problem for the reconstructions (less breaking but it was slicing wrong). It is better to use the header as the ground truth for what the shape should be, so I'm expanding the array out based on the expected shape, and using the header info to determine the channel index.