In the world of film sound post production, we have agreed on basic industry standards in order to “read” each other’s work. Every time you receive a video or audio tape, a CD, an EDL or an OMF file, these standards let you interpret them unambiguously. Standards have evolved over time, but they are as often based on arbitrary accidents, convenience or geo-

With a good understanding of the issues involved, and thorough planning, PAL should be no more difficult to deal with than NTSC.

politics as they are on what might be the most effective or obvious method. To complicate matters, competing standards have evolved to serve different countries.

One standard we have all agreed on is that there should be 24 frames of film per second of running time. Sadly, we seldom work directly with the film itself anymore — we use handier digital representations. Unfortunately for audio post people, film-to-video transfer messes up our tidy 24 fps world, because it creates several frame-rate standards (pulled-down and normal, NTSC and PAL). This forces us to make choices regarding playback speed at every step in post production.

I recently returned to California after working for four years at an excellent facility in Ireland called Ardmore Sound. While there, I had to make many adjustments, such as pronouncing “Beta” with a long “e.” I had to deal with zero as the first frame of action, instead of the head start mark. But the biggest adjustment was working with PAL, the 25 fps video broadcast format in use in most of the world outside of the U.S. and Japan.

Pulldown Basics

When transferring film to tape, two distinct operations are necessary, and both, misleadingly, are called “pulldown.” To distinguish between them, I’ll call the first “field padding,” and the second “speed varying.” When transferring from 24 fps film to NTSC video, both

Overseas television producers run film cameras at 25 fps, do their post production at 25 fps and deliver PAL. But features are more complicated because they’re shot and projected at 24 fps.

operations are used, while film-to-PAL transfers use either field padding or speed varying, but not both.

Let’s first review how NTSC works. What I call “field padding” is the calculation required to convert between 24 and 30 fps. The standard technique adds an extra video field (half-frame) to every other film frame (that’s the “3:2” in “3:2 pull-down”). This creates 60 fields (30 video frames) from the original 24 film frames, or an extra 12 fields per second. Surprisingly, the conversion is fairly transparent, because there are enough regularly spaced extra fields per second that they don’t normally catch the eye, except in areas of rapid motion. This process is used whenever film is transferred to NTSC tape, and since it faithfully preserves the speed and duration of the original film, we sound people tend to ignore it. (See How Do 24 Film Frames Equal 30 Video Frames?)

The second type of pulldown, “speed varying,” is also used during NTSC transfers. Black and white video ran at a true 30 fps, but NTSC color runs 0.1 percent slower —29.97 fps. To transfer film to tape, we have to slow it down by running it at 23.976 fps. To keep sound in sync with pulled-down film, it must also be slowed down 0.1 percent, but since this always happens, and there is no other way to work with video, we have become used to it.

Now, let’s look at PAL. Happily for television producers who work in countries other than the U.S. and Japan, you can easily run film cameras at 25 fps, transfer to video at 25 fps, edit picture and sound, mix, and do any final lab and video work, all at 25 fps, before broadcasting at (you guessed it) 25 fps. The complications arise when filmmakers shoot film at 24 fps, then

Working entirely at 25 fps and then pitch-shifting the printmaster can do real damage to sound quality.

do a film-to-tape transfer to PAL 25 fps video and expect to project in theaters at 24 fps (even European projectors run at 24). The great thing about PAL is that there is no need to deal with the subtle 0.1 percent speed adjustments we have to live with in the NTSC world. The sad thing is that the field padding needed to convert 24 fps film to 25 fps PAL looks terrible. We only need to add two fields per second, so every twelfth film frame is held for an extra field in transfer. This happens so infrequently that video appears jerky whenever there is movement in the shot. The alternative to field-padding is speed varying, but in PAL we must adjust by a whopping four percent, playing 24 fps film at 25 fps. This has a major impact on the pacing and running time of the film, as well as on the pitch of the sound. (A PAL Avid can create outputs either way. See Alan Stewart’s excellent article.) Both methods involve compromises and there must be careful advance planning and good communication with the post sound department to spell out the procedures that will be used at each stage of the post production process.

Good Alternatives

I’ll now describe what I consider the optimum method for getting production sound from DAT to Avid to ProTools in a PAL environment. Then I’ll cover some other common, but less desirable methods.

In practice, all dailies are telecined to PAL by running film at 25 fps and recording one frame of video for each frame of film. This is what a PAL Film Composer or LightWorks wants to see

The best way to get good tracks is for the picture department to load the sound digitally from the masters, sync it to picture in the Film Composer, and turn over via OMF.

as its input. The image is loaded silent into the Avid at 25 fps. Picture cutting is done at 24 fps in the Film Composer. The system does necessary field padding in real time (holding every 12th film frame for an extra field), so the image on a PAL monitor or in a PAL video tape output runs at 25 fps.

The best way to get good tracks is for the picture department to load the sound digitally from the masters into the Film Composer and sync it to picture, which will be running in the Avid at 24 fps. Then sound can get an OMF turnover. Assuming that production ran their cameras at 24 fps, all production sound will be a direct digital clone of the master, all editing done to the Avid output will be in sync with the film running at 24 fps, and the Music, ADR, Foley, etc. which are composed and recorded to the 24 fps video output will not be speed-varied or pitch-shifted at any time during post or exhibition.

Even if you don’t turn over via OMF, synching MOS picture and production sound in the Avid is the technique I prefer (and the one recommended by Avid). It allows the production of a frame-accurate EDL, referencing the original DAT timecode. The sound department can then reload the audio from the masters and autoconform using Titan (see my earlier articles) or PostConform.

If every picture department used this method, all would be well, but there are times when it is not workable. The most common problem is that because of telecine and transfer delays, the picture department can’t sync in the Avid and deliver dailies promptly. In addition, with a limited budget, the picture department may not want to use precious Avid time for syncing. Finally, OMF is still new to some people, and since there have been problems in the past, uneasiness remains in some quarters.

When the picture department does not want to sync in the Avid, there are two other good ways to work. In order of sound-department-friendliness, they are:

1. Sync in telecine and turn over via an EDL.
   
   
2. Sync on mag and turn over via an EDL.

Because the sound goes an extra generation or two, we can’t use OMF (except as a guide) with either of these methods. As we know, video transfer speeds up the 24 fps film to 25 fps, so to keep sync, the sound also has to be sped up by the same amount (4%). In the first of these methods, the production sound is loaded into a DAW (Sadie and Aaton are often used) with specialized software that does a simple 4% speed-up. The audio is pitched up in this process but will sound correct when slowed down to 24 fps in the Avid. The flex-file from the DAW is also loaded into the Avid, which can then keep track of the original sound timecode, so that an EDL can eventually be pulled for turnover. This EDL refers back to the original location recordings and will be in sync with a 24 fps field-padded video output from the Avid. The problem here is that the flex-file will not be perfectly accurate because the sound speed-up step isn’t frame locked, and autoconfomed sync can only be accurate to plus-or-minus two frames, at best. Nevertheless, this method works and has been the most common procedure, but it is slowly being superseded by the methods I described earlier.

The second method adds another generation but is still workable. Sound and timecode are transferred to three-track mag film and synced to the workprint for dailies projection in the traditional manner. The synced dailies are telecined together to video at 25 fps. A flex-file is made, containing the original DAT timecode from track three of the mag. The rest of the process is similar to method one, but because the timecode has gone two extra generations, neither of which is frame locked, errors will be larger.

Problems

There is one other method, which is common, but bad. European filmmakers usually work in TV and are comfortable with running everything at 25 fps throughout post production. When shooting a feature (at 24 or 25 fps) they often decide to finish everything, including scoring and mixing, at 25 fps, then pitch-shift up the printmaster 4% so that the composer’s intended pitch is preserved when the film is slowed down to 24 fps in projection. This can do real damage to sound quality, especially music and dialog — but it happens all too often. When confronted with this situation, I would generally recommend against attempting pitch correction — the solution is often worse than the problem. If you are forced into it, try using TC Electronic’s Studio 6000 multi-channel pitch shifter. It’s the least destructive device I’ve used. I’d also recommend individual pitch correction of each track of the stems before print mastering, if possible. Dialog may not require correction, as it could well be returning to its original pitch (if the show was shot at 24 fps). ADR will be another story.

Be sure to avoid “simul-DATs.” They’re effective for television-only releases, but when used for features, they can lead to serious juju. The sound is synced to video as in method one above, and simultaneously laid to the simul-DAT during telecine. The simul-DAT picks up timecode from video and makes it easier to turn over an accurate EDL, even without a flex-file. If production ran their cameras at 25 fps, these DATs will contain a clone of the master audio. But for shows that shoot at 24 fps, the audio on the simul-DATs will be sped up 4%, just like the audio on the videotape, and as a result, they’re no substitute for the original masters.

If you are compelled to use simul-DATs, you must get a 25 fps, sped up (rather than field padded) output from the PAL Avid. Your autocomformed simul-DATs will sync to this video, but the pitch will be 4% high. Music and ADR should be recorded to a 24 fps field-padded output. Dialog will be slowed down to its original speed in the 24 fps mix, but music and ADR will not be altered.

Conclusion

Most of us are familiar with U.S. video conventions and have made our peace with the intricacies of NTSC pulldown. PAL is another animal. It tends to cause editors and mixers plenty of heartburn when encountered on this continent. But with a good understanding of the issues involved, and thorough planning, PAL should be no more difficult to deal with than NTSC.