Implicit conversion from float -124.6840148698885 to int loses precision