Implicit conversion from float -247.07865168539325 to int loses precision