Implicit conversion from float -28.848167539267024 to int loses precision