Implicit conversion from float -12.0618556701031 to int loses precision