Implicit conversion from float -90.15463917525776 to int loses precision