What is non-linear post production? The term nowadays refers to computerised techniques of editing picture and sound. There is a plethora of devices on the market which perform these functions: Avid, Lightworks, DAR SoundStation, AudioFile, WaveFrame, Synclavier, to name but a few. They all do basically the same thing: those designed for picture editing 'digitise' picture rushes and offer highly flexible approaches to picture editing. On the sound side, the systems do much the same thing with location sound material in addition to handling all other aspects of post production sound work. The purpose of this article is not to debate the merits of competing systems or to argue the benefits of handling post production in this manner as opposed to doing it in more conventional ways. What I will attempt to set out are certain ground rules for productions contemplating using these techniques either for picture or sound or both.
Firstly let us consider what we are dealing with:
Camera original is video. Sound original is either married to the video image and /or is shot separately on a timecode format i.e. timecode DAT, Nagra 4 STC, Nagra-D. If the sound is not married to the video tape then the timecode DATs or 1/4 in tapes must bear the same timecode and user bits (which are commonly used to denote roll numbers) as the camera original otherwise the whole post production process is rendered impossibly cumbersome and expensive. Non-timecode formats are, in cost terms, not an option. So assuming that we are dealing with camera original that is married with sound, the location rushes are digitised into the Avid/Lightworks logged and then cut. An edit decision list is then produced which reflects every picture and sound cut the editor has chosen to make. Typically when viewed this EDL will show event numbers (i.e. the chronological order of the cuts), source roll numbers (contained in the user bits of the 80-bit timecode word), source in and out times (i.e. which chunk of the said location roll was used) and the destination timecodes (i.e. where the editor chose to put these chunks in the cutting copy). With the EDL is possible to 'on-line' edit the camera original and produce decent looking pictures that are an exact mirror of the often manky-looking (frequently digitised at low resolution to save memory) Avid/Lightworks cutting copy. It is also possible to auto-conform the sound from the camera original (or if available, the timecoded sound rushes) onto a digital sound workstation and hence considerably embellish the sound tracks. It goes without saying that the other sound elements apart from the original location audio can be added at this stage i.e. FX, Foley, ADR, music.
Camera original is 16 or 35mm negative. Sound is any analogue or digital sync format i.e. Nagra neopilot, FM pilot, Nagra timecode, Nagra-D, DAT timecode. Under these circumstances producers have a wealth of confusing choices.
Firstly, they want to cut picture and sound conventionally. This is not within the scope of this article.
Secondly, they want the picture non-linear and do the sound conventionally. This is not really an option as the sheer cost of producing reasonably current film dupes of a non-linear cutting copy from negative is prohibitive. Cutting room synchronisers that have been produced to allow accurate synchronisation between videotape and magnetic film have never really worked all that well.
Thirdly they want to cut both picture and sound using non-linear equipment. This is dead easy as far as 16mm post production is concerned since 16mm shot in the UK nowadays invariably runs at 25 fps and can therefore be easily scanned by any telecine array to produce true 25 fps video copies. The only issue here is how the sound is to be synchronised to picture. Since the whole picture cutting experience is to take place within the Avid/Lightworks it follows that the most economical way to do this is to transfer straight from the original negative to video tape. As far as the synching of the sound is concerned it is either done at the neg. telecine stage, or afterward to mute video tape rolls.
The principle disadvantages of synching sound to original negative are:
1. - You are spending expensive telecine time hunting for clapper boards.
2. - If the sound is shot on analogue timecode format there may be insufficient pre-roll time to lock picture and sound.
3. - If the sound is shot on non-timecode sync formats you will have to produce timecoded copies of the original sound rushes in addition to your later 'clones'.
4. - With the best will in the world telecine operators are not as skilled as second assistant editors at synching sound to picture. Errors at this stage are a considerable nuisance later on.
A far more efficient way to handle this process which works for all sync sound formats and gives you a fighting chance of getting rushes material that is in fact in sync is to re-record the location sound material into a low cost non-linear digital recorder, mark the claps, sync the latter to mute video rushes and when in sync, 'bounce' the sound across to the video and to 'clone' DATs' bearing the same timecode (and hence the user bits which identify roll numbers) as the videotape. The latter is now ready to digitise into the Avid/Lightworks and the whole process proceeds as described above, the only difference being that instead of conforming original video tape material to the non-linear cut you are now conforming film negative via a correlation between the time code carried on the video tape copies and the key numbers on the negative.
If you are shooting 35mm negative at 24 fps rate this process is considerably more tricky. To produce video tape copies of this material it is necessary to effect a 24 fps telecine transfer. This is because the PAL television standard insists on 25 television pictures per second. What is fact happens during the 24 fps telecine is that one frame, or two fields of picture are repeated every second. Unfortunately the location sound material cannot be made to sync up with these 24 fps telecines. Doubtless Avid/Lightworks will come up with a solution to this one as they apparently have for our US counterparts who are cursed with the NTSC 29.97 frame rate, Currently the only solution for those of us stuck with PAL equipment is to telecine the 24 fps negative at 25 frame rate, digitise the location sound, time warp it to a factor of 24/25, sync this to the mute video images and produce the clone DATs which will then auto-conform beautifully to the Avid/Lightworks cut and at the end of the day will, if replayed at 24 fps with a conformed print of the original negative, not only be in sync but will also be at the correct pitch. The process of time warping is something of a palaver since it has to be done in real time.
Lastly under Scenario 2, they want to cut the picture conventionally and do the sound non-linear. This is very much a favourite option for a variety of reasons. Firstly the sound has been accurately synchronised to picture and rubber numbered in the time honoured fashion. During the sound transfer process the original location time code is recorded on track three of the 35mm magnetic film alongside the audio on tracks one and two. Or if a non-timecode sync format was used on location, the transfer operator sets a timecode generator running with user bits reflecting roll numbers and records this timecode on track three as well as sending it, alongside the audio, to a timecode DAT machine, producing timecoded clones of the location sound material. The picture is then cut alongside the sync sound and thus track three of the cutting copy sound bears a flow of discontinuous timecode that can be read at the telecine stage to produce an EDL which when auto-conformed, provides an exact mirror image of the cutting copy soundtrack. Yes, a 24 fps telecine is used to provide video tape workprints for the sound editors but the final fruit of their labour is always in sync with the final cut picture negative.
This route has the added advantage that the cutting copy can always be shown to the director as a halfway decent image on the big screen. Obviously if having gone through the one-light print and conventional synching-up route, a decision was made to cut the picture using an Avid/Lightworks then these rushes rolls could be transferred via a 24 fps telecine to video, digitised and then cut. Providing that key numbers were accurately logged against the timecode on track three of the sound rolls then conforming a 35mm print for viewing purposes would be relatively easy.
There follows a brief description of how sound, assuming it is to be handled in a non-linear fashion, should be handled:
1.- Non timecode sync formats: timecoded DAT clones of the location sound rolls have to be produced at the rushes transfer stage. It is essential that the transfer is handled competently in terms of level and most importantly, the user bits generated in the timecode recorded on the magnetic film rolls and the clone DATs.
2. - Timecode sync formats: levels are really the issue here. A sound recordist lays down a line-up tone for a reason - make sure this is accurately carried across to the clone DATs and magnetic film.
3. - When making clone DATs from original material keep labelling clear and stick to one sample rate, either 44.1 or preferably 48kHz. Do not switch between the two.