You allude to the Kantian view that we should never treat humanity as a means only, but always as an end in itself (see Robert Johnson's explanation in the wonderful Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/entries/kant-moral/#8). Kant didn't think that we should never rely on people as means to our ends at all, but just not at the expense of regarding their humanity as an end in itself. Slavery is presumably a clear case on one side, and on the other might be a trade wherein both parties aim for mutual (not merely selfish) benefit. Whether, and in what cases, a user of pornography might be said to treat the model as an end in herself, is obviously complicated, but these considerations might well be relevant: Is the user aiming at benefitting the model? Is he adequately keeping in mind the effects on the model, on himself, and on the women with whom he will interact, as well as the effects on people generally of a culture in which pornography is prevalent?

The latter questions obviously point in an empirical direction: what are the effects of pornography on its models, on its users, and on others? Is it true that pornography use has psychological consequences that impact how its users tend to think about women, and how they intereact with women? See Caroline West's discussion, again in the Stanford Encyclopedia, http://plato.stanford.edu/entries/pornography-censorship/#5.1 . West's essay also addresses the rather different question of whether censorship laws might be justified.