Implicit conversion from float -53.070422535211264 to int loses precision