Implicit conversion from float -133.72881355932205 to int loses precision