Implicit conversion from float -111.86915887850466 to int loses precision