Implicit conversion from float -50.56603773584905 to int loses precision