Implicit conversion from float -34.874055415617136 to int loses precision