Implicit conversion from float -0.4241645244215988 to int loses precision