Implicit conversion from float -320.5882352941177 to int loses precision