Implicit conversion from float -12.806499261447556 to int loses precision