Implicit conversion from float -24.779005524861873 to int loses precision