Implicit conversion from float -16.54205607476635 to int loses precision