Implicit conversion from float -5.3602305475504295 to int loses precision