Implicit conversion from float -24.536741214057514 to int loses precision