Implicit conversion from float -84.1830065359477 to int loses precision