Implicit conversion from float -68.86699507389164 to int loses precision