Implicit conversion from float -57.93577981651376 to int loses precision