Implicit conversion from float -22.225475841874072 to int loses precision