Implicit conversion from float -8.595505617977523 to int loses precision