Implicit conversion from float -106.18528610354224 to int loses precision