Implicit conversion from float -62.3510971786834 to int loses precision