Implicit conversion from float -30.753138075313814 to int loses precision