Implicit conversion from float -83.42519685039369 to int loses precision