Implicit conversion from float -16.884681583476763 to int loses precision