Implicit conversion from float -29.666048237476815 to int loses precision