[Released] Terminator 2 CDS mix

[MrBrown]

* In addition, there is a Gaumont France / Europe DVD master that is basically the LD master with a Tri-Star logo in the opening, only in Pal format.

According to Rewind http://www.dvdcompare.net/comparisons/film.php?fid=160, there's early Australian and Brazilian DVDs put out by Sony that use the same Theatrical Cut LD master with the Tri-Star logo intact. The Brazilian DVD is said to be non-anamorphic and possibly a field blended PAL-NTSC conversion, which is odd. However, that's nowhere near as strange as Brazil using PAL-M (which is not the same as standard PAL, but basically the 525 line NTSC format with PALs 4.43361875 MHz color chroma) and having their media/video players accept NTSC content but convert their output a PAL-M 525 signal to a PAL-M TV.

The Australian DVD likely is the same as the early French (and Skandinavian) DVD (both not listed on dvdcompare). Same Subtitles, French Audio...

[MrBrown]

And never forget the first japanese Temrinator 2 DVD PIBF-1001.

I think it is also from the early LD master... also it is kind of odd, see the oddities thread.
https://fanrestore.com/thread-932-post-18062.html

[pipefan413]

You have four different masters to take into account for resyncs :

- LD + Early 1997 DVD the original T2 video master.

- Early Ultimate edition DVD from 2000 (might be in sync with the early master)

- X-treme up to 2015 BD they are all the same with just some differences in lead.

- Latest 3D degrain uprezzed master.

* In addition, there is a Gaumont France / Europe DVD master that is basically the LD master with a Tri-Star logo in the opening, only in Pal format.

This is extremely useful. Thank you.

Theoretically, if the Extreme DVD from 2003 is the same master as the 2003 Pioneer DVD (the Japanese one with the DTS track), that should mean that the DTS track from the 2003 JP DVD (the Pioneer one with the DTS 5.1 mix) should be possible to sync very easily indeed. I was under the impression more work might be needed because of differences in missing frames between the earlier versions, but perhaps that was only because the AC-3 version was off a 1997 disc (older master). This might be easier than I expected!

NARRATOR: it was not.

The remix audio from the TeamBlu release, which I think is just the 2008 JP BD, gradually gets further and further ahead of the DVD's CDS mix for some reason I can't quite figure out. Surely it's not missing more frames than the DVD... I took the centre channel off the TeamBlu 5.1 (TrueHD) and the DVD to check the waveforms and find clear-cut peaks to match, but the difference between them gets slightly bigger depending on how far into the file I look. And it isn't just after a reel change, I'm talking about it being a few samples out of sync again already after about a minute from the first adjustment etc. I'll need to keep investigating; I'm now trying the 2015 BD as a reference instead to see if it's any different at all.

The 2003 Extreme DVD also has a English Dolby Surround 2.0 track that is optimized for headphones, and is supposedly different than the 2.0 track (that Pipefan413 says is down-fold of the 5.1 EX mix on the JP Blu) on the 2000 Ultimate Edition DVD.

It does, so do some of the BDs (have headphone tracks) but I skipped all those for simplicity's sake (already a huge post). Was mostly focusing on the theatrical vs remix surround tracks. Determined to figure out a reasonably accurate resync of the DTS without doing a re-encode or decoding it at all, but seems that's less simple than I hoped.

[MrBrown]

...

Theoretically, if the Extreme DVD from 2003 is the same master as the 2003 Pioneer DVD (the Japanese one with the DTS track), that should mean that the DTS track from the 2003 JP DVD (the Pioneer one with the DTS 5.1 mix) should be possible to sync very easily indeed. I was under the impression more work might be needed because of differences in missing frames between the earlier versions, but perhaps that was only because the AC-3 version was off a 1997 disc (older master). This might be easier than I expected!

NARRATOR: it was not.

(The remix audio from the TeamBlu release, which I think is just the 2008 JP BD, gradually gets further and further ahead of the DVD's CDS mix for some reason I can't quite figure out. Surely it's not missing more frames than the DVD...)

...

Did you check, if both videos have the same Framerate?
Maybe the TeamBlu BD runs at 24 fps, while the NTSC DVD runs at 23,997 fps.

[pipefan413]

Did you check, if both videos have the same Framerate?
Maybe the TeamBlu BD runs at 24 fps, while the NTSC DVD runs at 23,997 fps.

Nah, TeamBlu is 24000/1001 (23.976)...

Also, turns out that the sync difference for the 2015 5.1 is the opposite way round, which is surprising. So JP DVD gets ahead of 2015 US BD (they start off in much closer sync) but 2008 JP BD gets ahead of JP DVD (and they start out much further apart in the first place). The 2015 one has seamless branching though so I'm wary of using it (eac3to reported that it was "skipping identical DTS frames"); the 2008 JP BD doesn't use branching because it's 2 separate BDs and yet it's the one that seems to be missing samples, as far as I can see. Having said that, it also isn't consistent across the whole file... it looks like if I sync the start chunk, the DVD gets ahead of the 2015 after a while, but at some point it flips around so that the BD is ahead of the DVD, then it flips back again... I can only assume that's reel changes differing but it might not be that simple (if frames are missing in random places or something). Deary me.

Something's weird here.


EDIT: If I sync the start of the JP DVD and 2015 BD by adding a small bit of silence to the DVD, it does seem to *mostly* (not 100%) remain in sync up until what I assume must be a reel change around the 18-minute mark. It goes from being sample-accurate to begin with to being about 9 ms out, then there's a motorcycle engine sound at approx. 00:18:05 and it's suddenly out by an additional 4 video frames' worth (4 / (24000/1001) * 1000 = 166 ms, plus the ~9 ms or so it was already slightly off by for some reason). So it does look like reel changes differ between the 2003 JP DVD and the 2015 BD (I think there are 4 frames missing on the DVD after the first reel change). Except... this only appears to be the case in the audio (they are different mixes, but it doesn't make sense to me that they would have significantly different sync by as much as 4-5 video frames when they seem to be released with video that doesn't differ at this section).

I wonder what audio source was used to record the theatrical ("CDS") 5.1 mix for the 2003 JP DVD and if it might have been missing frames that were therefore not recorded into the audio at this point, so when the remix was done later on from the actual masters, said missing audio was reincorporated...

Here's what the video looks like at the end of the motorcycle sound where it cuts away, in the JP DVD version and then the TeamBlu tweak of what I reckon is the JP BD:





I'm now more determined than ever to get hold of this old French BD with the theatrical mix on it so I can see how that one's synced... if anybody has it ripped already, or is able to rip it without too much of a faff, please let me know. I can't find the damn thing anywhere.

[Stamper]

I have the french disc, no problem.

I think the Team Blu releases have the tracks hacked off to keep in sync, it was that way for The Terminator.

However be warned, when I synced old The Terminator LDs to the current master, I had to fill in the gaps between reels either by extended both sides to cover a missing gap, or using another source.

So if you want to sync anything on a project like T2, you would need to go back to the original released track, and not use any release by anyone else.

[pipefan413]

I have the french disc, no problem.

I think the Team Blu releases have the tracks hacked off to keep in sync, it was that way for The Terminator.

However be warned, when I synced old The Terminator LDs to the current master, I had to fill in the gaps between reels either by extended both sides to cover a missing gap, or using another source.

So if you want to sync anything on a project like T2, you would need to go back to the original released track, and not use any release by anyone else.

Well, I don't have the original 2008 Japanese BD but as far as I can tell that seems to be what TeamBlu based their T2 on. I'd much rather be looking at the original Japanese BD if I could! I'm just about to check if they did anything to jonno's sync of the 1997 US DVD 5.1; afaik he synced to the UK Skynet Edition rather than the JP disc I think TeamBlu's video is from, though they might already be exactly frame matched. So far, it does actually look like they've at least hacked off the start with a negative delay (looks like -817ms, a little over 20 video frames) but I dunno yet if there's any further tinkering.

That said, that's why I went back to the 2015 BD, which I do actually have now (recently imported it). And that's also not in sync with the JP DVD regardless of the offset at the start. If I add 2 video frames' worth of silence to the start of the JP DVD (effectively, +83ms, though I'm doing it more precisely than that to look at the waveforms with minimal error for now so it's actually 4,004 samples) then it remains basically in sync with the 2015 BD's remix audio up until that point I just mentioned where it cuts from Arnie on the bike to Sarah in the facility.

I'm wondering if any of this inaccuracy is being caused by the branching / editing to allow the Special Edition bits to be added into the theatrical cut; I know that this old French disc and the Japanese one (which are both from 2008) are 2-disc BDs with SE on one disc and TC on the other so I wish I could import them... but I don't want to spend ~£100 importing the JP one and the only place I could find the FR one was also nearly £100 and I don't even think they post to here.

[Stamper]

The Arnold on Bike, cut to to Sarah, is the very moment the new remaster is off sync with the 2015 version, incidently.

I can provide most audio from all the versions untouched. (DVDs, BR)

[spaceboy710]

So excited to see this project come along!

[pipefan413]

The Arnold on Bike, cut to to Sarah, is the very moment the new remaster is off sync with the 2015 version, incidently.

I can provide most audio from all the versions untouched. (DVDs, BR)

Interesting. Do you know if that is indeed a reel change? It would make sense time-wise I suppose, reels are generally a little under 20 mins afaik.

That would be marvellous, although if anybody else can chip in to save you re-ripping stuff, that'd be great. (Do gimme a shout if you're reading this and can provide any of these, especially the unmodified 2008 FR BD 5.1 English DTS-HD MA, to save Stamper a job!)

In the meantime, I'm testing whether I can (very very carefully and slowly) go through the waveform checking where DTS frames can be inserted or knocked out where necessary to achieve a more accurate sync than was previously possible without decoding. It would've been nice if it were as simple as just applying a delay based on it being the same master, and it might sound perfectly reasonable to do that, but it wouldn't be that accurate. Both of the other syncs I've got here don't line up all that well with the 2015 BD regardless of delay, largely because of the approx 166 ms discrepancy around the 18m 5s mark I mentioned. I think if I gradually insert carefully located DTS frames (which are in this case 512 samples long, or about 10.666... ms in terms of duration) I can have it catch up before it gets to the 18m 5s bit, and then I'll test a few approaches to see what works best for lining up the bike sound with the visual cut. Theoretically, you might expect that bike sound should end at precisely the end of that video frame (it doesn't fade out, it's a hard cut in both the video and audio). So...

If you count from 0 as QuickTime does, the last frame of Arnie On Bike is frame number 26038 (i.e. the 26039th frame). To convert to audio terms, that's:

video frames / video frame rate x audio sample rate

26039 / (24000/1001) * 48000 = 52,130,078‬ audio samples by the end of that video frame
(or to make it simpler to do loads of calculations, 1 frame = 2002 samples, so 26039 x 2002)

In the actual 2015 (remixed) audio track, the bike sound *actually* ends with sample #52,130,134 and ...135 is the cut. That's only a 52,130,134 - 52,130,078‬ = 56 sample discrepancy, so clearly that sound is indeed intended to land right on the video frame cut.

I'm working in DTS frames, and to sync the beginning of the audio pretty closely I can insert 6 DTS frames (e.g. 6 x 512 = 3072 samples) at the start of the file but if I make no further changes then the sync starts to drift slightly as time passes and then drastically changes at the bike-to-Sarah cut, which happens on sample #52,123,580 (52,130,078 - 52,123,580 = 6,498 samples, almost 13 whole DTS frames, earlier than it theoretically should). That's a discrepancy of >128 ms, which is not insignificant especially on a hard cut. So I'll try to line up everything before that moment and see what still needs to be done to fix the bike-to-Sarah cut, methinks.

All of this is made harder by that pesky 16 kHz background hum, which means "silence" is not actually silence at all and inserting any real silence would be jarring. Instead, I'm going to have to sample out a 512-sample chunk (1 DTS frame) of nothing but 16kHz hum and insert that for sync fixes.

EDIT: This is coming along nicely so far but it's slow work. Just discovered the first place that the 2015 BD (remix) actually *removes* time rather than adding it: the "get down" gunshot. It seems that it isn't just the SFX for the gunshot (I thought maybe it was replaced), but there's time removed just *before* the gunshot. You can see what I mean in the following helpfully labelled waveform screenshot (which is from around 00:27:39-00:27:40), I accidentally cropped out the timestamps):



There's about 3.5 DTS frames (roughly 1,752 samples, 36.5 ms, a little less than 1 video frame's worth) of sound here in the theatrical mix that is apparently missing from the remix which throws it behind the remix/2015 BD sync for a good while. What's weird though is that almost the exact same amount of audio is then later added back in (some at 01:00:45 when they're in the car being chased, a good chunk later at 01:01:11.100 or so when it cuts away from the pursuer to them high-tailing it away and we get another engine noise, this one starting much later on the remix than on the theatrical cut).