Implicit conversion from float -24.775725593667545 to int loses precision