Implicit conversion from float -29.101123595505626 to int loses precision