Implicit conversion from float -15.693950177935946 to int loses precision