Probably not quite what you had in mind, but because I'm half way around the world from my teammates the most successful tools I've used are a console editor and tmux (potentially running on a VM if you don't trust your friend). For voice, mumble is my first choice because it is both low bandwidth and low latency. Tiger VNC works reasonably well to display a shared web browser. Nothing else I've tried has the fidelity and low latency to do serious pair programming for long periods of time when one person is really far away.
Having said that, almost anything works well if you are close to each other or if you are only going to pair for an hour or so. Even non-collaborative conference call-type tools work fine. One person does some work. Then they check in their code to a git repository. The other person merges the changes, shares the screen and picks up from there. It means about a 1 minute "keyboard handoff", but it's pretty reasonable.
For you, though, I think the main thing is trying to find a solution that allows you to capture video and audio from both parties easily. You don't want to be relying on two parties to record and then editing because it's pretty risky if something goes wrong. I would first look at mutli-gamer streaming setups and then start seeing how a collaborative session might be done.