Implicit conversion from float -34.03100775193798 to int loses precision