Implicit conversion from float -16.484641638225256 to int loses precision