Implicit conversion from float -3.0343511450381726 to int loses precision