Implicit conversion from float -18.11320754716982 to int loses precision