Implicit conversion from float -94.97757847533634 to int loses precision