Implicit conversion from float -46.85446009389673 to int loses precision