Implicit conversion from float -81.60493827160494 to int loses precision