Implicit conversion from float -10.491573033707851 to int loses precision