Implicit conversion from float -120.80097087378641 to int loses precision