Mask files are supposed to follow the logic that values of 0 (or NaN/NULL) are excluded and any other number marks a pixel that is to be included. Byte data type images behave differently; only pixel values that are equal to 1 are includes -- pixel values from 2 to 255 are excluded.
If a file has been filtered with a pixelmask, then users must only ever bin the image or event list (table) at the resolution of that pixel mask. This includes attempting to bin and filter in the same command.
Attempting increase the bin size of the image or event file results in an inconsistenty between the mask pixels and the data included in the table or image. This will cause tools such as dmextract to compute the wrong area and/or to compute the counts incorrectly.
Users should bin their pixel mask to the desired final size before using it as a filter.
When working with event files, users should bin the mask first, bin the event file into an image at the same resolution, and then filter the image with the mask.
Users who wish to combine event files (ie observations) which have been filtered with pixel masks should carefully inspect the output prior to use.
The information about the pixel mask filters is stored in the file's subspace. As files are combined with dmmerge, the subspace is merged. When the files have multiple difference, the result of this merger may yield multiple subspace components; only the first of which is used by tools like dmextract when computing the area.
How and when these multiple subspace components are created is not always predictable; therefore users should carefully inspect the merged output to be sure they are getting the expected results.