Implicit conversion from float -8.423326133909285 to int loses precision