Implicit conversion from float -90.51575931232091 to int loses precision