Implicit conversion from float -44.07643312101911 to int loses precision