Implicit conversion from float -48.595041322314046 to int loses precision