Implicit conversion from float -12.416918429003033 to int loses precision