Implicit conversion from float -44.80176211453744 to int loses precision