Implicit conversion from float -45.816062176165815 to int loses precision