Implicit conversion from float -1.1054637865311179 to int loses precision