Implicit conversion from float -46.49068322981367 to int loses precision