Implicit conversion from float -9.406779661016955 to int loses precision