Implicit conversion from float -8.025477707006374 to int loses precision