Implicit conversion from float -70.45296167247386 to int loses precision