Implicit conversion from float -102.96482412060303 to int loses precision