Implicit conversion from float -8.062953995157386 to int loses precision