Implicit conversion from float -48.549618320610705 to int loses precision