Implicit conversion from float -5.290581162324656 to int loses precision