Implicit conversion from float -22.755798090040926 to int loses precision