Implicit conversion from float -0.3162055335968432 to int loses precision