On the problem of the horopter