Implicit conversion from float -1.1392405063291164 to int loses precision