Implicit conversion from float -16.866952789699567 to int loses precision