Implicit conversion from float -380.33057851239664 to int loses precision