Implicit conversion from float -16.900878293600996 to int loses precision