Implicit conversion from float -54.055482166446495 to int loses precision