Implicit conversion from float -17.203389830508485 to int loses precision