Implicit conversion from float -53.17180616740089 to int loses precision