Implicit conversion from float -30.801526717557252 to int loses precision