Implicit conversion from float -14.803493449781655 to int loses precision