The actual recording of the dialogue went pretty well. The setup of the ADR lounge is actually pretty nice; it's separated from the control room of the studio by enough distance that there is no coupling of the sounds. There is a big-ass video monitor in there, and speakers so that I can play a DVD from in the control room, and the victim in the ADR booth can see and hear the scene that they're recording. I can set the DVD player on repeat so that it loops over and over, and the victim can rehearse. Then the sound gets shut off, and the recording begins.
Most of the time, it's good within the first four takes. It actually goes pretty fast when it's operating, it's the prep that takes all the time (like identifying the scenes, burniong them to DVD, figuring who needs to be in what scene, etc.), as well as the post-recording part which is the most tedious because it involves a lot of trimming and massaging and manipulating.
I spent a chunk of time last night doing the tedious part of the dialogue replacement, which is the fitting of the recorded dubbed dialogue to the image. All I did was one scene, just the dialogue (no effects, no background sound, nothing else), and it took a little over two hours for about a one-minute scene. And it's really amazing that even those dubbed lines that look like they were dead-on while recording really need to be manipulated quite a bit to make them match when you see it over and over. Even if it's just a little bit off, it looks wrong. VocAlign helps, but I ended up only being able to use it about a third of the time; the rest of the matching was me doing it by hand. I'm a bit surprised that I was able to get better matches by hand in most instances-- I've done it before, but apparently I've done it enough to be pretty good at it.
I sent that finished dialogue piece to avindair as a test so that he can insert it into the edit and see if we can effectively match up on both sides of the internet. It's not a good long-term solution because these files are big; I think the sound file alone was 11Mb. Even with both sides having broadband, that's a frustratingly slow method of transferral.
The ADR is happening in two stages: the first stage is a quick-and-dirty replacement of dialogue that has technical faults so bad as to be unintelligible; things like traffic noise, places where we just couldn't get a microphone close enough, extraneous noises like coughing, or places where we just could not record sound at all. As much as I would really love to be able to use all of the dialogue that was recorded on set, sometimes it just doesn't happen that way (particularly with location shooting). The second part will happen after Marscon, which will involve more in-depth replacement and smoothing of the sounds, trying to make them as unobtrusive as possible. It's a real judgement call, because the characteristics of the recordings are so different and you can't really have them both in the same scene (or more precisely the same scene section) without yanking the audience right out of the movie.
It's sort of like the first round of the ADR is triage. It's patching the gaping wounds so that the patient doesn't bleed to death. The second phase is more like cosmetic surgery, making it look and sound better, more aesthetically pleasing. It involves more things liek "tuning" the sound to the space it's in with things like reverb and equalization, adding Foley and effects, and balancing the whole sound space.
There were some definite things that came up that I need to change, really some pretty small technical details. But overall, it worked out pretty damn good.