Implicit conversion from float -24.47963800904978 to int loses precision