Implicit conversion from float -16.409395973154375 to int loses precision