Implicit conversion from float -41.546572934973625 to int loses precision