Implicit conversion from float -156.51810584958216 to int loses precision