Implicit conversion from float -46.06299212598424 to int loses precision