Implicit conversion from float -55.75471698113208 to int loses precision