Implicit conversion from float -6.18320610687023 to int loses precision