Implicit conversion from float -56.61654135338344 to int loses precision