Implicit conversion from float -22.63250883392226 to int loses precision