Implicit conversion from float -145.45081967213116 to int loses precision