Implicit conversion from float -135.2199413489736 to int loses precision