Implicit conversion from float -33.425196850393704 to int loses precision