Implicit conversion from float -23.04926764314247 to int loses precision