Implicit conversion from float -14.746423927178142 to int loses precision