Implicit conversion from float -55.9349593495935 to int loses precision