Implicit conversion from float -35.160771704180064 to int loses precision