Implicit conversion from float -190.5194805194805 to int loses precision