Babe Pig in the city
The latest appearance of Australia's favourite porker mixes innocence, sophistication and Hollywood's latest ingenuity. Richard Buskin talks to the men behind the sound of a family of animals in full flight
by Richard Buskin
'The first hazard for the returning hero is fame,' states the narrator at the start of Babe: Pig in the City. In bringing our porky pal from the land Down Under back to the big screen, director George Miller - who produced and cowrote the original - seems to have pulled out all of the stops.
Accordingly, no fewer than 799 animals appear in the new Babe, which sees the champion shepherding pig attempting to save his and everyone else's bacon after Farmer Hoggett falls into a well and the farm is plunged into bankruptcy. Babe and Mrs Hoggett set out for the state fair in order to earn some much needed prize money, yet, when she is arrested en route for supposedly smuggling drugs, the two of them are stranded in a fictitious metropolis that incorporates such notable landmarks as the Statue of Liberty, the Sydney Opera House and the Hollywood sign. So it is that their less-than-childlike adventure begins, using four times as much animatronic and computer-generated special effects as the original movie, and, in so doing, nearly tripling the production costs.
'The whole film was a challenge,' says Steve Burgess, a highly experienced freelance engineer who took care of the Foley work at Sound Firm in Melbourne, while the actual filming took place at Twentieth Century Fox Studios in Sydney. 'I don't think I've ever worked on a film that was harder.
'A major reason was that it featured hardly any people and there was no sync sound. We were often creating sounds that were bigger than the real-life sounds, and we were always trying to heighten the track, so nothing was straightforward.'
For its part, Sound Firm invested in a DSP Poststation system in early-1998, having evaluated editing alternatives such as Pro Tools, Fairlight and Sonic Solutions over an 18-month period. 'DSP were the most responsive in terms of addressing the needs of the whole facility,' says in-house engineer Ralph Ortner. 'We sometimes have up to ten editors with as many as four different projects running at the same time, and so it is important for us to be able to mix and match systems, make smaller or larger systems, transfer tracks to backup, get tracks from the editing stage down to the sound mixing stage.
'At the time that we approached DSP they hadn't completely solved a lot of those issues, but they were very open to input from us to help develop a systems approach to postproduction. We wanted something that wasn't just an editor, wasn't just a mixer, but started to bring those worlds together a bit more and made the transition between the two a lot easier. DSP are very progressive and what impressed me was their response to our needs. A lot of the computer editing systems have been designed by computer boffins without really understanding the postproduction process, but DSP spent a lot of time actually sitting and watching how we work in order to understand what we want. So, rather than us conforming to their particular software they ensured that their software would meet our needs.'
The result is a 4-station setup comprising a fully automated 32-track Poststation, with a control surface featuring touch-sensitive faders, an 18Gb local memory and NLV (Non-Linear Video); a 24-track desktop system with NLV that is used principally for Foley and ADR; a 16-track desktop system that is used for ADR as well as sound effects and dialogue editing; and an 8-track desktop system that is used for effects editing. A central server, the DSP Team, connects the four stations together, and has 36Gb of memory providing 41/2 days of continuous recording.
'This means that for a particular project we can have somebody recording Foley on one system, another person recording ADR, someone doing sound effects editing, and still be mixing,' says Ortner. 'So, you can be doing pickups or changes as the project is progressing, instead of transferring to a backup tape, and then unloading from a machine and carrying across. It's really an integrated network, and that's been fantastic for the sort of high-end episodic television work that I've been mixing.'
'The DSP is a far easier system to use than Pro Tools,' adds Steve Burgess. 'Ifound Pro Tools to be a little too awkward for a studio setup when you're recording live. I mean, with the DSP system you literally have the 24 tracks in front of you running in time, and you're recording straight into that track as you would on a multi or any other system; it looks and works more like an old multitrack recorder. At the same time I suppose the other great advantage is the ability to splice, cut and move sounds very quickly and very easily.
'There again, when comparing the DSP to the Fairlight, the first time that I jumped on the DSP system I didn't like the way the recording function would happen on the cue-like start position. That's because I like to have the microphone open at different degrees depending on what we are doing. I phoned them up and said, "Look, this isn't the way Iwant it to operate", and the next morning when I walked into work there was an email with an attachment that provided me with a wider parameter of opening it earlier or shutting it or having it on the line. Generally their backup has been fantastic. Every time I've asked for something it pops up within a day, not weeks or months.
'A major part of what I like about the product is the NLV. With the hard disk drive I can scroll across and look at everything. It has a cueing system in it which gives you a lot of variables; you can have streamers, beats, counts, and everything changes according to the way you want it to function, and that's especially important when you're working with actors. Some actors might like a cue beat, some might not. They all have their different tastes, and it's nice to be able to sit there and change things instead of getting into an argument with them. In fact, the system initially didn't have cue beats, so again I rang them up during the day and the next day I had an email with the appropriate software.'
For Babe's Foley work, Burgess was using the aforementioned 24-track desktop system, boasting 30Gb each of both audio and video memory, together with a Yamaha 02R. 'The DSP is absolutely fantastic for recording Foley,' he asserts. 'It's also great for ADR, and extremely handy in a mix room where you run it as your playback machine. We have a Harrison Series 12 here, and rather than use the 02R to do my mixdowns I lock the DSP up to the Harrison, and, if certain sounds are a little out of sync or there's something that I want to move or grab from somewhere else, I can even move it on the screen and continue to mix.'
Sound Firm's post work on Babe: Pig in the City started in July of 1998 and ended in November. Roger Savage took care of the rerecording, and, being that a large proportion of the voices was emanating from the mouths of animals, he certainly had more than enough looping to keep him busy. Nevertheless, given the nature of the animatronics work, it was very much a case of the digital effects people fitting the lip movements to the dialogue rather than the other way around.
'I think this film provides a bench-mark for the future in terms of the ability to now change a dialogue line after the "actor" has been filmed,' says Savage. 'You change the line and then you change the mouth. I think that will eventually be applied a lot to live action with humans, because it will give the director the freedom to change lines without having to worry about how good the lip-syncing is.'
Steve Burgess, meanwhile, worked alongside Foley artist Gerry Long, while Craig Carter reconformed their Foley to the different versions and assistant Andrew Neil took care of backup support in order to keep everything running.
'Babe's Foley track required quite a lot of detail in it, and we ended up recording about 3,500 clips of sound,' confirms Burgess. 'You see, we were looking at a film that had no sync track to it, and George's requirements amounted to a high degree of accuracy. A good example is the sound of the dogs' feet - because they are four-legged animals and they were running so fast, it was impossible to record all four feet at once and maintain a high degree of accuracy. So, what we did was record a guide track with Gerry just tapping his fingers in sync with the dogs' front feet, and a second track tapping his fingers to the back feet, and I then cut these two tracks on the DSP against the image, tightening them up and getting the rhythm right. I noticed that if you slide the front versus the back you can change the rhythm, and it was great with the DSP because I could easily toggle on the hard drive. We were literally cutting down to a half-frame accuracy or less.
'Next we would record the pads of the dogs' feet. George needed the dogs to feel heavy and big, and the sound to be strong, and in order to put that much weight into the sound we couldn't do it in sync at all, because it was way too fast. So, for the pads, we would do one track first using either sandbags or boxing gloves or sometimes Gerry running on the balls of his feet, and I'd feed him the guide track of the front feet that we had done with the finger-taps and he would then try to stay in sync with it. That, of course, was impossible, so we would always varispeed and then I'd cut those pads back in time to the guide track against the front feet and then against the back feet.
'On top of that, at times we wanted to get a fleshier sound, so in those cases we would use boxing gloves or even pads of hands to get some slap and put that up against the pads and the guide tracks. Then there would be another two tracks for the toenails, again doing back and front, so literally every time a dog moved we would have two guide tracks for the taps, two tracks for the heavy-type feet, two tracks of the fleshy-type feet and two tracks of the toenails; eight tracks running every time a dog moved. What I found was that by varying the levels in association with each track we could get a lot of good motion in there, and we really needed to do that in order to get it to work. A lot of the time, when I couldn't get the weight that was required, I used the DSP's varispeed and then its Dilate function; I'd varispeed it and slow it down, and then dilate it back to its original length. I'd make each individual footstep a clip and then position it on top of the template.'
While the dogs' feet serve as an example of the degree of sync quality that was required, every single sound effect in the film necessitated an enormous amount of effort. Sample something seemingly as simple as the rattling sound of the dogs' chain - if Burgess and Long thought that a 12-foot chain with 12-inch links would suffice they were wrong; it actually sounded more like a crate of milk bottles.
'George really wanted to hear those links clunking together, but if you really grab a chain and move it there is not much attack in the signal on each link,' Burgess explains. So, what to do? Well, once again 16 tracks came in handy to create the desired effect courtesy of a hefty chain, some pulleys and winches all being ground together, along with a metal bar being run along a cast iron grate. On the other hand, the sound of the duck's feet were altogether easier to attain.
'All we used were two kids' flippers, and it was a one-element pass every time,' says Burgess. 'However, when the duck would fly away there were 500 wing-flaps left and right, and for that we had to cut every single wing- flap, each running five to ten frames in length, and checker-board them in order to be able to mix them. Again, at certain times we would use two different types of feathers and vary the levels in order to help achieve the movement of the bird.
'Because we'd have three dogs running in a scene I'd end up using 32 tracks on the DSP just to create the dogs' feet. So on every spool I had three or four premixes to bring down, and we were averaging between 60 and 100 tracks per reel. That is far too difficult to mix all at once, especially with a Foley reel where you have a sound happening every two seconds or less. Eventually I supplied the final mixes on the dubbing stage with a 16-track split of the Foley.'
In the end, around 400 hours were spent on recording and between 150 to 200 hours on the mix. Roger Savage did the main mix on a Harrison console at the Sound Firm facility located on the Fox lot in Sydney, and the film was finalled there using various formats; DA-88, Akai and Sony digital dubbers, 2-inch multitrack and Pro Tools. Meanwhile, just to add to Steve Burgess' workload, there were also around 200 hours spent on reconforming.
'If we were locking off reels and finishing they would cut the picture again,' says Burgess. 'So we were consistently reconforming to the new pictures. I've never worked so many hours before on a film, but George Miller was adamant that he wanted the best possible results out of it... We were running Sony STD 9000 data backups, and at night I would have my assistant Andrew backup everything that I had done during the day while I would walk into the other room and keep mixing until 4 am. I tell you, I was averaging between 80 to 100 hours a week.
'On the first day of the mix George actually rang us here, and he just said, 'The work is absolutely bloody brilliant.' That was Reel 1, and after he'd said that I then had to keep the quality up there until the end. The pressure was really on.'
Studio Sound Jan 1999
Visit FilmSound.org - Free Learning Space for Film Sound Design