Implicit conversion from float -83.65145228215766 to int loses precision