Implicit conversion from float -0.05424954792043479 to int loses precision