A plugin-filter providing a highpass command. It subtracts the Gaussian blurred input image from the input image.
If “Shift Values to Display Range” is unchecked, the offset, i.e. the output created by flat portions of the image, will be 0 for 32-bit float, 32768 for 16-bit and 128 for 8-bit (grayscale or RGB) images.
If the full image is processed, for grayscale 8-bit and 16-bit images, the grayscale (pixel value) calibration will be set to have zero value at this level. Thus, one can use, e.g., Process>Math>Square after high-pass filtering to highlight all pixels deviating from their surrounding.
Note that Undo will revert only the image contents, not the calibration.
High_pass.java”; uppercase/lowercase matters.
— Michael Schmid 2007/06/28 15:12