Implicit conversion from float -27.341772151898738 to int loses precision