Implicit conversion from float -82.45989304812835 to int loses precision