Tiling correction when image size is not multiple of tile size

[ArneDefauw]

Tiling correction fails when one of the image axes is not a multiple of tile size. This is caused by this code:

        # create the masks for inpainting
        i_mask = (
            np.block(
                [
                    list(tiles[i : i + (ic.shape[1] // tile_size)])
                    for i in range(0, len(tiles), ic.shape[1] // tile_size)
                ]
            ).astype(np.uint16)
            == 0
        )

This could be a problem if user cropped the image before running sparrow.

[lopollar]

go to bigger crop for tiling correction, then takethe crop again to visualize

[ArneDefauw]

It is difficult to implement a satisfactory fix for this.

• going to bigger crop is impossible, due to the assumption the user already cropped the image before starting sparrow pipeline.
• we could do padding (via reflect), but for this we would need to know where the stitching lines are (which we do not know anymore due to cropping ).
• we could go to smaller image, but then again we would need to know where the stitching lines are.

Won't fix for now, as it seems an edge case scenario, added a check if image dimension is a multiple of tile size, code will now raise a ValueError with a clear error message.