Implicit conversion from float -145.44303797468356 to int loses precision