Implicit conversion from float -135.54744525547446 to int loses precision