Implicit conversion from float -50.40920716112532 to int loses precision