Implicit conversion from float -16.39810426540285 to int loses precision