Cost-Effective Alternatives to Synchronized-Sound Dailies

Cost-Effective Alternatives to Synchronized-Sound Dailies

by
Evan T. Chen
May 4, 1999

An edited version was published in the October 1999 issue of
TV Technology
Copyright © 1999. All Rights Reserved.

Table of Contents

  1. Introduction
  2. Data Transactions
  3. Not an Easy Task
  4. Conquering Ratios
  5. Best Method
  6. Conclusion


Introduction

Video dailies for digital non-linear editing figures one of the largest and seemingly insurmountable costs in filmmaking. For 35mm to one-light Beta-SP or 3/4" sync-sound dailies -- the two most common formats of choice for low-budget independent filmmaking, although everyone should push for DV and its assorted "pro" variations now -- most laboratories charge per hour on a guaranteed 4:1 ratio cap at no less than $275/hr. To make it more appealing these labs will frequently fudge the numbers and "lower" the rate, say, to $220/hr at 5:1 or $220/1000', which is really no change at all. What all these numbers mean is that 1 hour of film footage to tape with timecode and window burn will cost roughly $1100 + $45 stock + $??? hidden costs, such as a one-time setup charge for framing, timecode, and burn-in, so we'll just assume $1150. At a 10:1 printing ratio, for a 2-hour feature that's roughly 20 hours worth of telecine at $1150/hr which amounts to $23,000. Another way to look at it is think production: 4500'/day at 35 days is 157,500', and assuming 125,000' printed, at $220/1000' that's $27,500. Either way a low-budget filmmaker will spend roughly $25,000 on sync sound dailies, not an insignificant amount of change.

Telecine MOS (soundless) dailies halves the time at the lab, a surefire method of reducing transfer costs of this hypothetical project by $12,500, but now how does one synchronize the production sound? This article presents three alternative, cost-effective approaches to synchronizing production sound to MOS dailies based on all-digital transfers, benefiting from the advantages inherent to reproducing digital content, like multiple lossless generations and the convenience of not having to align or calibrate machines. Unfortunately these methods also suffer from digital audio's confusing sampling rate synchronization issues, so much of this article will be devoted to this aspect. Return to the Table of Contents.


Data Transactions

All-digital transfers are especially important because of today's OMF (Open Media Framework) capabilities, common in many audio and video workstations, which allow the interchange of data and metadata. For movie projects, this means that all of the preparatory work created by the Picture Editor, including temp music, VO, ADR, sound effects, dialogue editing, and so forth, can be handily exchanged with the sound editors. Although digital media never quite interchanges this smoothly in practice, most producers and facilities still prefer this means over tediously and expensively re-creating what's already been done, particularly the time and expense saved in re-transferring and autoconforming and re-synching the location sound via a pull-list or EDL. However, having the laboratory create sync-sound dailies onto analog video tape, like the usual Beta-SP or 3/4", immediately constitutes two generation losses in audio: one in creating the dailies, a second recording into the non-linear picture editing workstation. Hence, via OMF, the Dialogue Editor receives audio material already tarnished. After predubbing, finalling, printmastering, and an optical track, that's a whopping six generation losses(!), a substantially noticeable difference from the original recordings, although in the final release the other audio elements -- sound effects, music, and so forth -- help mask the degradation. From this example, digitally recording and transferring audio has a tremendous qualitative impact on the final outcome.

Ideally, this is a best-case scenario. What happens more often is that dialogue editors receive OMF audio production elements that were originally transferred way too hot into an Avid with all kinds of hum and buzz from the 3/4" dailies. That's an additional $12,500 for the lab, plus the salary to whomever poorly transferred the audio, for material that eventually will be incorporated into the film two generation losses down, distorted, and buzzy, not to mention the additional week necessary for the dialogue editor to clean up otherwise useable material, another week's worth of ADR to replace the distortion, another week's worth of foley to cover the holes from the ADR, and another week's worth of mixing time to manage all of these additional elements now -- easily another whopping $25,000! What usually happens? "We can't afford that... (but we're creative!)" So in the end, audio post gets no time to create miracles (hey, even God had six days) and the clients end up with a hack-n-slash sound job that nobody's satisfied with, a no-win situation neither party should have agreed to in the first place. Now of course I'm exaggerating a bit, but only slightly. Return to the Table of Contents.


Not an Easy Task

So, who is to synchronize the production sound? Well, assuming that a consumer DAT machine is available, typically the Assistant Editor is assigned the tedious task of digitally transferring and syncing to the MOS picture dailies overnight on the non-linear picture editing workstation. Even at a union scale of $1225/week over the presumed 7-week shoot, that's $8,575, still a savings of almost $4000. It's rare however to have an Assistant Editor know so much about post-sound synchronization issues to accomplish this with minimal risk. It's a complex subject that only a handful truly understand, even in the audio post community, so the finer points will be briefly discussed below.

The dangers involved with this approach primarily revolve around the digital non-linear editing system configuration. For example, certain machines have only frame accurate sound, or are accurate to only a sub-division thereof like half-frame or quarter-frame: few allow audio nudging by the sample, although this will probably change soon with newer software and hardware releases. Since sound editors depend on the dailies for audio sync accuracy, any errors made in synchronizing sound will be on the order of these whole frames or respective fractions, compounding the localized sync discrepancy already inherent in the 3:2 telecine pull-down process (very few of the many articles discussing the 3:2 implementation cover its effects on audio sync and its implications on sound post, another involved topic all together), although cutting on a true 24-frame platform trivializes this point. There are some companies that have developed specialized software to compensate for the 3:2 discrepancy, like generating a 24-frame cutlist from a 30-frame video EDL. These customized databases manage telecine transfer field/frame relationship information and therefore can adjust timecoded audio accordingly. Unfortunately, audio autoconforming is not ideal either: the robust but old RS-422 protocol, the primary standard used for transport control over all types of video and audio decks, is also not sample accurate, nor is LTC (longitudinal timecode) and definitely not MTC (MIDI timecode, used by many digital audio workstations now, is about half as accurate as the other formats), although some hope still resides in VITC. The technology is there, however, for sample-accurate digital transfers, and as of this writing I know of one innovative company that's created a surprisingly affordable piece of hardware that allows just that -- unfortunately they've marketed it more towards the music production side of things, like most digital audio workstations used in post nowadays.

What if the editing platform doesn't support digital transfers? What if an analog Nagra were used on location? If so, I strongly suggest hiring a specialist for a day to configure the setup properly for in-sync analog transfers. One can risk having the Assistant Editor transfer and sync things analog, but I can't begin to stress how complicated the issues are in synchronization and properly matching gain structure (balanced to unbalanced connections, cabling, headroom, calibration, alignment, etc.). Those that say otherwise just aren't aware of them... yet (hey man, ignorance is bliss). Furthermore, an expensive, professional DAT deck with timecode and/or external sync capabilities should be used under these circumstances, which may cause one to reconsider having the laboratory provide sync-sound dailies since the savings now may not be worth the risk. A consumer DAT player may be substituted, and in all probability the sync will be adequate as long as there aren't any 5-minute long takes, but this varies depending on the model of the DAT machine and editing equipment, so again it's chancy. Nagra's require even more setup and more know-how, including resolving pilot tones and aligning the machine and the like, and the production will have to rent one with its associated synchronization gear, which is again pushing the limits of the budget. Forget about purchasing an analog Nagra: even used they're expensive and they're being used less and less now. Return to the Table of Contents.


Conquering Ratios

My final remark on this approach revolves around sampling rates and timecode rates. In order for a digital transfer to be done accurately for editing purposes, the digital non-linear editing system must be able to handle the production sound's sampling rate to timecode rate ratio. For example, if the production sound was shot at 48k:30ndF (ndF = non-drop frame timecode) which most productions are, then the editing platform must be able to handle 48k:30ndF or an equivalent ratio, such as 47,952:29.97ndF. Certain non-linear machines, as well as particular high-end digital video decks, can only handle 48k:29.97. Audio transferred digitally into one of these machines will drift one frame every 30 seconds, unless the location sound mixer knows beforehand to shoot everything at 48.048k:30, the more "fashionable" but less familiar production sampling rate to use nowadays. This special sampling rate to timecode rate ratio also allows an all digital path, from sound editing to digital mixing to layback to a long-form color-corrected digital deck, without additional sampling-rate conversions in between, which basically never happens on any feature, regardless of budget. Ironically with every boutique touting "all digital" this-and-that nowadays, I don't know of one film that's had an all digital audio path. I know, some facilities claim to "choose" mag or 2" over digital recording, but the fact of the matter is few are truly equipped for all digital.

An alternative to the Assistant Picture Editor syncing sound dailies is to have the audio post house sync the dailies. These facilities should have the know-how and necessary equipment already calibrated, aligned, gain-matched ready to go, including sample-accurate workstations, expendables, timecode DAT machines, 1/4" (Nagra compatible) mono and stereo machines equipped with center-track timecode and/or pilot, and an audio-insertable Beta-SP deck. The money saved won't be as much as the previous method but the production still ends up paying only around two-thirds the estimated laboratory fees, about $8500 dollars or so, billed on an hourly basis. The trade-off is in the expertise, but often the piece of mind and passing the responsibility off is worth it.

There are several advantages to this approach. It's great to have an experienced individual in audio post listening to production tracks early on who may be able to spot and prevent future problems, as well as suggest wild lines and/or effects to shoot on location. Furthermore, digital audio workstations are sample accurate, so any mistakes will be on the order of samples, not quarter-frames or half-frames or God forbid whole-frames. Audio workstations also handle a wider variety of sampling rates and timecode formats than their picture-editing counterparts, so more than likely the production sound will be transferred digitally (if recorded digitally to begin with). An audio boutique has higher-grade A/D and D/A converters for those analog recordings. The equipment is cleaned, well-maintained, aligned and calibrated, and gain-matched for optimal transfers. Everything is resolved to house-sync, ensuring accurate transfers as current synchronization technology allows. The post house bares any technical burdens on their part. And when sound editing rolls around, all of the production dialog is already transferred and ready to be edited.

The disadvantages are numerous too though. Few audio post houses seldom are in the tip top shape I mentioned above, so some investigating is necessary. Some may not have an audio-insertable Beta-SP deck, so one will need to be rented, typically for around $750/week. At seven weeks however, the rental house should accommodate a better deal closer to $500 or even $400 for about $2800 total. The post facility will undoubtedly have a tough time dealing with the alignment of the audio and timecode tracks on the Beta-SP deck, an absolutely crucial step in this process, since it is a new issue for them and does require some specialized thinking. The turn-around time will undoubtedly be higher too. Expect an additional-day delay, and keep in mind a means of transporting the dailies. Also, although having an audio boutique sync dailies is a novel idea that benefits all parties, it's a relatively new concept that'll be a first for many facilities, and like all new concepts, skepticism, uncertainty, and reluctance often greet them. Return to the Table of Contents.


Best Method

As a final option, the best one in my opinion, rather than assigning an Assistant Editor or someone novice to the task, I highly recommend hiring a specialist or consultant to develop a tailored approach particular to the situation and handle the syncing process autonomously. As I've pointed out, the pitfalls using this approach are many, but a specialist should have the background and experience necessary to deal with the problems I briefly mentioned as well as ones I didn't all the while saving costly laboratory fees. At roughly 45 minutes of footage per day for a low-budget independent feature film, it shouldn't take longer than 3 hours. Even at $50/hour, that's no more than $750/week, which over seven weeks is $5,250, a savings of $7,250! Return to the Table of Contents.


Conclusion

It's never too soon to prepare and think far ahead, even about post-post-production before pre-pre-production. An approach towards post-production should be chosen prior to beginning production since it directly influences decisions in production and will effect the quality of the project. I must stress again however that it requires a thorough understanding of the myriad of technical, aesthetic, and logistical facets from the beginning to end in the filmmaking process to successfully implement cost-effective techniques. Yes, if implemented correctly, thousands of dollars can be saved. If implemented incorrectly, it'll cost twice that to salvage it. Hire a specialist to handle this aspect during production. You'll come out ahead financially, technically, and aesthetically. Return to the Table of Contents.

Evan T. Chen freelances as a Post-Production Consultant, Sound Supervisor, and Sound Editor in California and New York. He worked most recently on "The General's Daughter," "South Park," "Mystery Men," and the independent feature film "Restless." Feel free to contact him at evantchen@yahoo.com.

[Note: this was the original biographical blurb that appeard in the 1999 issue of
TV Technology - Evan 2/21/03]

Copyright © 2010 by Evan T. Chen. All rights reserved.