Implicit conversion from float -100.94955489614244 to int loses precision