Pink Floyd - Pompeii - Restored 1974 Cut
 
Album info
 
Recording Date  : Various
     
Length  : 84:44
Format  : DVD
Track List
 
01 Main Movie 84:44

Notes
Recorded at the Amphitheatre, Pompeii, Italy. The original quad mix synched with the movie. Excellent.


From the info file:

PINK
FLOYD.
POMPEII
Restored 1974 Cut

Audio Lineage
35mm Cinemascope 4-track magnetic striped print > custom-built playback machine > Digidesign 192 I/O > Pro Tools HD > 24-bit/192kHz WAV > Pro Tools HD > Izotope SRC > 24-bit/96kHz WAV

> Final Cut Pro > Compressor > 16-bit/48kHz AC3 > DVD Studio Pro > DVD Video

> SurCode MLP > discWelder Chrome > DVD Audio

Video Lineage
PAL Director’s Cut DVD > MPEG Streamclip > MOV

NTSC Vestron LaserDisc > Pioneer LD-S9 Player > Datavideo TBC-1000 > Canopus ADVC-110 > AVI > MOV

> Cinema Tools > Final Cut Pro > Compressor > M2V > DVD Studio Pro > DVD Video


Source Notes

The audio was taken from a 35mm Cinemascope 4-track magnetic striped print of Live at Pompeii featuring the quadraphonic mix (L, R, Ls, Rs; '4 corners'). This format used the same 4 channel magnetic audio tracks but not in the intended standard L, C, R, S configuration. This '4 corners' quad format was briefly used in the early/mid 1970's for more music focused films. The only theaters equipped for the non-standard audio format were in New York and Los Angeles.

The audio on this print is severely corrupted and the evidence in the recording suggests it was the printmaster the production copy was transferred from that was flawed from the start.

The video was taken from the 'Original Concert Film' featured on the Director’s Cut DVD release, as well as the Vestron LaserDisc release. The DVD is of decent quality, as it displays some occasional compression artifacts (including macro-blocking and mosquito noise). The LaserDisc, on the other hand, shows signs of laser rot in the form of the multi-colored specs that appear in the video. Color correction was applied to the LaserDisc footage to make it look as good as possible.


Historical Background

In 1974, associate producer Phil Borack (whose company April Fools Films handled distribution of the film in the United States) and director Adrian Maben visited Los Angeles for the purpose of finding someone to produce a quadraphonic mix for the film from the original multi-tracks. Dan Wallin, who received an Oscar nomination four years earlier for his work on Woodstock, was appointed for this task. The mix was done in discrete quad, the multi-tracks were mixed to mag at A&M record studios, then the mags were taken over to Todd-AO where the final mix was done.

The quad mix was featured in select screenings of the film, but has never seen the light of day since then. Pink Floyd was never involved with the quad mix, and it's possible that they’re completely unaware of its existence.


Technical Background

Production of film with sound involves the audio and visual content being separated during phases of production and joined back together at the end. This involves recording time code (basically a clock tick) on both parts so they can be matched back up later. The difficulty of doing this with analog equipment requires skill and attention to detail and even then the equipment can go out of sync and require multiple attempts.

Here are some excerpts from a document posted on filmsound.org (from the section titled: 'My mix sounded great on the mixing stage - but my print isn't in sync!'):

'If the Final Mix Printmaster has been transferred or copied, be sure the copy was done correctly. We have had experiences where a perfectly fine Printmaster was thrown out of sync because a copy was made first, and the optical shot from the copy.'

'If the Mag Printmaster was in sync when you reviewed the final mix, check to be sure the film lab didn't accidentally ‘misprint’ the soundtrack by moving the optical negative a perf or two, or a frame or two when they married it to the picture. This can easily happen IF THE HEAD POP or TAIL POP is not EXACTLY CORRECT on your final Printmaster.'

'If you printmastered in 2000-foot film reels, and FOR ANY REASON these reels were then separated and rejoined later, this poses a prime opportunity for sync to slip. If the beginning of a 2000 foot reel is in sync, and the last 1000 feet is suddenly (and consistently) out of sync until the end of the reel, suspect this phenomenon immediately.'

The instruction is: 'You should IMMEDIATELY contact your post sound house and tell them what you've experienced.' These instructions were apparently not followed for this print.


The 35mm Print Used

I can only guess whether or not there were added difficulties due to the different implementation of the 4 audio tracks. Whatever happened, it appears this 35mm film was produced from a flawed printmaster.

The 4 channels of the audio on this copy are out of sync with each other. More accurately, they are offset from each other. If slid back to their correct position, the tracks match up in sync with no ebb and flow. The offsets range from .5ms to 160ms and the channels are offset from each other by different amounts in different sections. The end result is a entirely unlistenable clattering train wreck. The 4 channel quad audio has clearly been printed to this copy 1 channel at a time and in different sections. Channels 3 & 4 have more breaks and sections independent of channels 1 & 2 with channel 4 also having a few more than channel 3. The audio in channels 3 and 4 sounds noticeably inferior to the front channels. Easily the sound of generation loss. Channel 4 is a narrower tape strip than the other 3 channels (a compromise deemed a moot point when used for a single rear fx channel) and is noticeably grainier than channel 3 but 3 & 4 still sound like a generation loss vs channel 1 & 2 quality. This film went through some combination of editing and reassembly and having audio redubbed with what sounds like maybe a final reducing but only for the front channels (leaving the rear channels a little more spliced up and higher generation). Some of the break points correspond with points where the performance was edited (some songs were edited from multiple performances).

I'm speculating that the printmaster is at fault because there is evidence that identifies this 35mm print as being copied from an already flawed master. There is crosstalk recorded on this copy of the film. Some channel 2 audio into channel 4 and some channel 1 audio into channel 3. This can happen if the head or tape alignment is off enough for signal aimed at one track to bleed into the other. The 2 > 4 crosstalk is the worst for the 1st reel. This is the only really noticeable part unless you were to listen intently to the Rs channel by itself. The very small amount of 1 > 3 crosstalk is only perceivable during the dialog sections where the rear channels are silent. You hear the 2 > 4 crosstalk in these sections as well. This crosstalk was recorded from an already offset master. The offset audio in channel 4 comes in later in time than in channel 2. With channel 2's audio bleeding into channel 4, the result is a 'pre echo' when you listen to the corrupted parts of channel 4. But wait, there's more. Additionally there were 3 sections of channel 4 that were reversed polarity. The waveform screen shots include examples of all this.


Audio Remastering Notes

There were enough audio elements present in multiple channels in enough places to determine the correct position to restore the sync between tracks. Suddenly the clattering train wreck turned into an amazing Pink Floyd surround sound mix!

Now the main issues were:
1. A piercing saturation, especially in the cymbals, hitting at about 5kHz.
2. A healthy amount of tape hiss along with a 12kHz whine.
3. Some short missing sections from the beginning of reels 1 & 3.

I dynamically attenuated the saturation with the Universal Audio Multiband Compressor. I reduced the hiss and 12k with the Izotope RX2 Denoiser by different amounts depending on the program. The recording was still very “toppy”. I brought up everything below that 5kHz saturation point to restore the proper balance. Finally, since there's always some loss from tampering (especially noise reduction), the original audio was blended back in enough to restore nuance without sacrificing adding back too much noise.

There was an audible pre-delay in a few spots in channel 4 from the crosstalk and offset issue. Mostly just the 1st reel in isolated spots in Echoes part 1. The worst artifacts that were very audible were carefully edited out to restore to original content. The parts of the interviews that are dialog only are mixed mono in front. The originally silent rear channels now contained crosstalk during these sections. This has been silenced during the dialog only parts. (Note that for the offset corrected only pre-master, these sections will still contain this crosstalk which will be noticeably offset.)

The heartbeats intro was cut in 2 places. A section at 1:41.9 - 1:46.7 was missing as well as a short section from 2:05.6 - 2:06.5. The 1st 1:41.9 section (with leader) was spooled up on a 6th reel. The remainder of reel 1 started from the 1:46.7 mark. The 2nd missing section was found spliced together creating a jump in time in the audio. The 1st 1.5 seconds of Rick's piano were missing from the top of reel 3. These missing sections were recreated by recycling nearby parts. We are very fortunate that only such easy to fix (possible to fix) pieces were missing! Drastic dropouts and noises were repaired along the way.

Finally, level corrections were made. Channel 2 was considerably hotter than the other 3 and more so for the 1st 2 reels. Correct channel balance was determined by a combination of noting max channel levels and observing surround panning consistency in the mixes as well as comparing obviously intended balances to the stereo and mono mixes. The level drops on all the loudest sections of the songs likely as a result of some compression or limiting originally done in post production to conform to the more limited dynamic range of the film's audio media. This was corrected.

This is intended as a conservative mastering job to repair or remove the most distracting flaws. Compromise was in too little noise removal rather than in losing any quality by trying to hit a more professional standard. This recording could still likely be made to professional release standards with more time and delicate scrubbing. Some purists that would rather hear hiss than any trace of program damage might argue that nothing more is needed. The pre-master is intended for anyone who would want to remaster starting from the unprocessed audio.


Acknowledgments

Thanks to those who helped with the project:

Arnold_Layne for conducting the LaserDisc transfer.
Joseph Christ for overlooking the transfer of the quad mix.
Jimfisheye for remastering the audio.

Special thanks to the respective individuals who donated and helped fund the project.