Implicit conversion from float -30.69970845481049 to int loses precision