Implicit conversion from float -10.557324840764338 to int loses precision