Implicit conversion from float -74.85981308411215 to int loses precision