Implicit conversion from float -48.75562218890556 to int loses precision