Implicit conversion from float -71.57556270096464 to int loses precision