Implicit conversion from float -33.10690423162585 to int loses precision