Implicit conversion from float -291.7177914110429 to int loses precision