Implicit conversion from float -16.792144026186577 to int loses precision