When a source itself is too bright, there will be a depression at and near the peak position of the source in the image proceesed in the standard way, because the pixels around the peak are so heavily piled up that most or all the events are detected with a bad grade (such as, Pattern 36 in MOSs) and have been filtered out before the image or event is created. In extreme cases those pixels may not contain hardly any event, hence are not regarded as piled-up pixels by this task, since the count per frame in those pixels would not exceed the threshold.
However even in those cases this task should work nicely, because (1) somewhere away from the peak position of the source, there must be pixels which are not as badly piled-up as the peak position, but show a large enough the count per frame rate to be recognised as piled-up, (2) the OOT event streaks are then defined based on these `peripheral' pixels, (3) the area of those recognised piled-up pixels should be distributed in a roughly concentric area from the source, therefore they overall mask out the large chunk of area, which includes the source itself. Even in the case a source is so bright and causes so much pile-up that some entire rows in a CCD chip are not recognised as piled-up, it will not matter practically, because any of those pixels will not be recognised as a source in the (standard) source detection, hence there is practically no need to mask those positions, providing the mask file is used in conjunction with the result of a source detection.
XMM-Newton SOC/SSC -- 2016-02-01