Implicit conversion from float -25.41916167664671 to int loses precision