Implicit conversion from float -78.41584158415841 to int loses precision