Implicit conversion from float -10.481430536451171 to int loses precision