Implicit conversion from float -52.118226600985224 to int loses precision