Implicit conversion from float -56.131805157593135 to int loses precision