Implicit conversion from float -22.11901306240928 to int loses precision