Implicit conversion from float -63.5128805620609 to int loses precision