Implicit conversion from float -84.59514170040487 to int loses precision