Implicit conversion from float -108.43049327354262 to int loses precision