Implicit conversion from float -27.262357414448672 to int loses precision