Implicit conversion from float -16.371191135734065 to int loses precision