Implicit conversion from float -14.628099173553721 to int loses precision