Implicit conversion from float -27.94258373205743 to int loses precision