Implicit conversion from float -44.669338677354716 to int loses precision