Implicit conversion from float -0.7239819004524861 to int loses precision