Implicit conversion from float -156.4968152866242 to int loses precision