Implicit conversion from float -14.04624277456648 to int loses precision