Thought Experiments: Modern PCM Audio sync'd to Film

[Ryan Gallagher]

Starting a thread to collect this topic. Due to Vista Vision efforts, we've seen at least a couple approaches to modern or alternative digital sound sync'd to analog film for presentation (not just in the studio for dailies). Some approaches perhaps more generally applicable to all prints than others. Perhaps there are some other approaches? Just a place to think out loud.

Prior relevant threads where the topic was broached...

VistaVision Returns
https://www.film-tech.com/vbb/forum/...vision-returns

One Battle After Another, and the film-is-God press
https://www.film-tech.com/vbb/forum/...m-is-god-press

Out in the wild:

1. An electro mechanical sync generating a dual-phased timecode. I believe this was the WB/Steve Levy/BLS approach with the TCM Fest Vista Vision (and perhaps One Battle as well for the Boston screenings). A Shaft Encoder fed an AVID Sync HD unit, allowing ProTools 5.1/7.1 mix to sync to dailies or a screening. Although we never did hear what their "backup" audio approach was, although it was stated on the Perf Damage podcast they did have something in mind for a backup at least.

Downside... seems to rely on frame perfect presentations and prints, with no interruptions, cause regaining sync after a restart seems problematic. Missing frames of vintage prints could maybe be "indexed" ahead of time in a timecode translation layer, but restarts still seem fairly insurmountable to me, unless you had other markings on a print to represent valid restart framing locations?


2. One Battle VV prints were printed with DTS timecode in an alternate track location using modified 35mm DTS readers. It seems there was a variation in approaches depending on screening location? Boston being PCM, and others perhaps playing a DTS track per "normal"?

Downside... although a new track location and stretched to the 8perf format, this seems to be just DTS audio, not modern PCM, unless there was some clever use of other supported track formats towards a lossless encoding.


3. With the addition of a Dolby Renderer, it's speculated that even syncing ATMOS should be possible at least with the dual-phased encoder sync approach, and all that is really lacking is someone wanting to do it for a special VV/70mm screening etc.


Here are my initial continuing thoughts to throw into the mix:

In a perfect world a PCM 5.1/7.1/ATMOS mix source would actually be a DCP as it is now, since booths are already set up for that without additional hardware, and to better utilize other things tied to DCP playback such as CCAP etc etc.

So how do you sync one? Is it possible? With any methods not already described?

Based on the gear in my booth, I would be tempted to experiment with the ShowVault/IMB "Chase Mode" via SMPTE timecode coming into SDI inputs. SDI chase mode seems to have been a factory option (for multiple server/projector arrangements), and only requires a valid embedded SMPTE timecode that I can tell.
https://kb.dolby.com/CinemaKnowledge...-ShowVault-IMB

There are also LTC methods using a separate feature stack on some dolphin models (if ordered with such options soldered on).
https://kb.dolby.com/CinemaKnowledge...-ShowVault-IMB

The crux would be a translation layer between proprietary printed DTS timecode and readers or simplistic encoder based dual-phase timecode, and getting it embedded as non-locked slew-able SMPTE video timecode onto an SDI line. I expect a hardware or software translation layer would have to be somewhat invented, and also one is at risk of infringing on DTS/Datasat's IP by even attempting it.

I'm not as familiar, but DTS processors also supported SMPTE timecode, perhaps someone can enlighten, but I believe the feature is to play back to an incoming SMPTE timecode, not that it supports reading two different printed timecodes on the film. XD-10 manual references TN-H171 article for more information on SMPTE playback, but that technical note seems to have evaporated off the internet. If in fact it supported reading SMPTE off a print we'd be getting somewhere. DTS "Special Venue" information also talks about playing back when there is no time code printed, using the tachometer to generate SMPTE, perhaps useful as well if one was able to pull out that signal somehow.

To make it really painless what you really need is a time machine, and go back and have a DTS engineer spin up an XD-10 option card that outputs SMPTE tiemcode based on the incoming DTS print timecode. Would only be valid for DTS having prints, but that is something!

Having never used IMB/server syncing methods, I'm also unsure if the "following" machines would satisfactorily freewheel or chase as needed for missing footage or non-perfect frame rates of the projector. The feature name "Chase Mode" however implies they might do so, rather than sync once at startup and never resync.

Downsides: This may not be as generic as the encoder dual-phased timecode to SyncHD approach, cause I expect each vendor/system would have to be explored for similar feature of chasing of an incoming timecode. But it perhaps has the advantage of not needing a ton of extra kit in the booth, sound processor does it's thing and foist the following timecode onto an existing feature of projectors/servers?

In the end, perhaps no DCP syncing methods available are adequate to deal with the larger variances in a mechanical projector, and it's a bogus idea, but it's a fun thought.

[Dave Macaulay]

The problem with an encoder is faced in IMAX film projection. Missing film means sync lost after that point if frames are cut out. A damaged print needs to have black film spliced in to replace the damage until a replacement reel is available, and even then black frames may need to be added at the splices if they can't be respliced without any loss.

[Ryan Gallagher]

The problem with an encoder is faced in IMAX film projection. Missing film means sync lost after that point if frames are cut out. A damaged print needs to have black film spliced in to replace the damage until a replacement reel is available, and even then black frames may need to be added at the splices if they can't be respliced without any loss.

Yeah the encoder method seems fine for one-offs... but not robust enough for release prints to lots of venues over time. IMAX had their procedures for frame loss as you suggest, but those would be hard to retrain into your average 35/70mm booth. Whether you are splicing in black, or creating a detailed catalog of all missing frames for software workarounds, both are problematic in practice.

That is why translating DTS (or replacing it with something else), a frame/time aware code, seems favorable over the dual-phased pulse code, if that is even possible.

Encoder method could perhaps be made more robust if black repairs were normalized and a modern printing had a edge code that included frame counts periodically, one could re-thread to a known frame number, adjust a software/translation device start-up point, and restart a presentation.

Edit: Such an edge code would also permit you to "check" prior repairs. Without it you are just trusting.

[Ryan Gallagher]

Here is a post-shower crazy thought. DTS processors can all display the timecode on their LCD. So just point a camera at the LCD and do a bit of real-time OCR on the video feed, then in software you could translate that into actual SMPTE, LTC or whatever you need.

Not all timecodes or implementations permit chasing/slewing... even if you were relegated to generating a dual phased pulse and the SyncHD+Protools methods, the OCR software side could perhaps detect time jumps and slew until caught up etc.

It would certainly be potentially easier to recover from an interruption than a shaft encoder, cause the timecode source is still DTS, although a pulsed output could not tell the system where to pickup, but the OCR side might be able to determine that part, and perform a jump resync before pulsed resume.

So crazy it might actually work. lol Not a robust solution, but a fun one.

Admittedly the camera is a bit silly, there are probably ways just to intercept either the LCD characters directly, or more appropriately a timecode bitstream somewhere in the XD10. My background is more software than hardware... hence the bolt on OCR thought.

[Carsten Kurz]

The signal coming from the DTS reader is easy to decode. Some DTS units even have options for SMPTE LTC. There is a SMPTE LTC chip on the DTS-6D DTS-TC board.

However, the print needs to have a DTS TC.

Bildschirmfoto 2025-11-13 um 04.47.14.png

[Josh Jones]

If the DTS timecode is already printed, this seems like the only logical way without reinventing the wheel. Assuming that the current generation players can do LPCM 5.1/7.1, just ensure that conventional DTS material is available for older players.

Also, doesn't DTS spit the running time out on a serial port somewhere? I know Dolby sends block numbers.

Carsten, apparently we posted at the same time.

[Leo Enticknap]

The problem with an encoder is faced in IMAX film projection. Missing film means sync lost after that point if frames are cut out. A damaged print needs to have black film spliced in to replace the damage until a replacement reel is available, and even then black frames may need to be added at the splices if they can't be respliced without any loss.

That problem goes back to Vitaphone! The problem back then was that some projectionists were more detail oriented than others, replacing one or two damaged frames with black in order to ensure that perfect sync with the disc was maintained. Others could not be bothered, believing that no-one would notice a frame or two's sync loss.

However, in those days, the loss of an odd couple of frames because of an unsuccessful attempt to make a cement splice (you scraped too deep), or perf damage, was very common. Polyester is tougher stuff than nitrate, added to which today's film prints are not used as intensively as 1920s prints were (three day booking playing four times a day > crossover to another theater > three day booking > rinse and repeat for six months), and so accidental loss of frames is not the risk now that it was then.

[Ryan Gallagher]

Also, doesn't DTS spit the running time out on a serial port somewhere? I know Dolby sends block numbers.

That bit seems handy, I didn't look into the 6D manual for SMPTE information as Carsten shared. Is running time serial out present on the XD10/20 as well? I didn't spot that in the manual when searching for SMPTE, but if is just a copy of the run time counter I might have skipped right past it. Will have to look at see what serial is doing during a film or test sometime. If decoding the DTS readers directly is easy as Carsten says, then one is half way there anyway.

One could maybe stick with DTS native methods if all you want is LPCM/DTS-MA 5.1 or 6.0 via alternate ST files (if supported and generated). But I was really trying to imagine the full monty, various methods for syncing to 7.1/ATMOS or other future even more modern STs to a print. Or 70mm DBOX screening anyone?

I know you can set up SMPTE and perhaps the other special venue functions via the XD10/20 screen menus, it just wasn't clear in manual on if it was telling it to read printed SMPTE, respond to external SMPTE TC, or output the TC somewhere, or some combination of those things.

It appears had I turned my head and fully read all the information on the timecode offset page at the back of the XD10 manual it would have been illuminating. It supports actually reading both DTS and SMPTE printed timecodes. If it was easier to implement with SMPTE TC on a print, that would require special venue printings, which might be higher hill to climb.

Timecode_Offsets.png


I have a broadcast video card that supports per frame RP-188, VITC, or LTC timecode manipulation via the SDK, so doing a partial slew/chase test with ShowVault/IMB in SDI Chase Mode wouldn't be that difficult. Perhaps something for a rainy day.

SRD is definitely more prevalent in 35mm, I hadn't even considered if there might be a block number based approach for 35mm. Is it really a block ID number, or just a read block count (Like does the number jump when there is missing footage)?

On the missing frames subject. I'm trying to imagine something that works for both new releases but also legacy prints that have been around the block and are already missing frames. Running mostly back catalog prints, my impression of how much a problem missing frames would become might differ from a new release venue.

[Marco Giustini]

well if all you need is LPCM 5.1, then the XD10/20 should already support it natively. How was that called? Coherent Acoustic? I believe it was up to 8ch of uncompressed sound - but only supported by the XD10 and later players. I believe I did ONE screening with that format and that was it ?

Now, whether the existing workflow has the knowledge and expertise to generate that format, I do not know.

[Sean McKinnon]

I know you can set up SMPTE and perhaps the other special venue functions via the XD10/20 screen menus, it just wasn't clear in manual on if it was telling it to read printed SMPTE, respond to external SMPTE TC, or output the TC somewhere, or some combination of those things.

Its even simpler than that Ryan... No special version DTS player needed.

[Ryan Gallagher]

well if all you need is LPCM 5.1, then the XD10/20 should already support it natively. How was that called? Coherent Acoustic? I believe it was up to 8ch of uncompressed sound - but only supported by the XD10 and later players.

Yeah was reading about that too. Not sure about 8ch, I know smpte special venue was for 6 channels originally, , I did not catch if this lossless alternative was capable of 7.1

[Ryan Gallagher]

Its even simpler than that Ryan... No special version DTS player needed.

Reading between the lines it sounds like you’ve tried this. ;-)

[Marco Giustini]

Yeah was reading about that too. Not sure about 8ch, I know smpte special venue was for 6 channels originally, , I did not catch if this lossless alternative was capable of 7.1

It's been a while but I think I showed Sony's "Big Fish" and it was 7.1 - that is, 5 screen channels.

[Ryan Gallagher]

It's been a while but I think I showed Sony's "Big Fish" and it was 7.1 - that is, 5 screen channels.

For sure, but 5 screen channels and contemporary 7.1 are two different animals. XD10 provides for 8-10 channels of output but the intent is for recreating the classic widescreen 70mm audio screen channels, and not necessarily doing a modern 7.1, although I'm sure if one wanted to author a 7.1 8ch DTS or Coherent Acoustics tracking and do some remapping of the outputs into a 7.1 room, seems possible.

Here are the XD10 v2.1 test mode channel sequence and output mapping for example.

channel sequence.pngchannel output.png



Getting to 7.1 (with LRS and RRS) with DTS, even lossless, seems doable if people wanted to deviate from a few norms. Would only be valid on new releases or remasterings. But syncing ANY type of format available via a sync'd DCP seems more of an interesting thought experiment. Like maybe there is an ATMOS or IMB tracking of 2001, or maybe there should be, that could be re-imagined with the existing 70mm DTS prints for venues that can do immersive formats as well as film? Even if the prints are missing footage???

My original quip that started this whole rabbit hole had to do with directors putting in all this effort for this huge array of formats (including film) in modern releases, often favoring film as the "intended" in their talking points, but leaving the "best" audio experiences to the exclusively the digital only venues. Would be nice to have the audio cake and eat it too if they are going to tell audiences their best experience is with 70mm or VistaVision etc.

[Marco Giustini]

Well, even in the CP650 and the CP750 the two extra channels are labelled Le and Re (e for "Extra"). They're channels. You can route them wherever you want. It's still a 7.1 setup ?