Implicit conversion from float -145.64954682779458 to int loses precision