Implicit conversion from float -5.0664136622391 to int loses precision