Implicit conversion from float -56.56160458452723 to int loses precision