See this Comm Center thread by Arto Oksanen. This came up because he was using an IEEE float image yet PP reported a saturated star yet it was only 14000 ADU. Conversion to integer internal to PP may be off scale!
The Comm Center thread has data and script to reproduce the problem.