Implicit conversion from float -0.7547169811320771 to int loses precision