Implicit conversion from float -8.267716535433067 to int loses precision