Implicit conversion from float -16.423357664233578 to int loses precision