Implicit conversion from float -12.209944751381215 to int loses precision