Implicit conversion from float -70.75819672131146 to int loses precision