Implicit conversion from float -12.006507592190886 to int loses precision