[Released] Terminator 2 CDS mix

[pipefan413]

I just got the Brazilian Columbia Tri-Star R4 NTSC DVD in an auction , so we can find out what it's DD 5.1 track is all about. It should be here in the next few days. 

BTW, has anyone heard or have any information on the AC-3 LD [LD68952-2DD]? Is it based on the CDS mix as well?
Since AC-3 LD tracks are know to sound better than DVD ones, could it be a better source?

What about it's PCM track, is it based on the regular 35mm Dolby Surround track?

I don't know anything much about T2 LDs myself, I've only really read up on DVD onwards and most of that info came from others here e.g. this very thread (I just went away and researched further from there). If you're not already aware of lddb.com, that's handy up to a certain point but will not necessarily be helpful in identifying which particular mix is present (I take it you're referring to this disc).

-

Here's what I think I'll do re. T2:

1. Clear some damn disk space somehow so I can actually work properly. Also: finish Snowpiercer so I can clear out the WIP files for that (and probably delete some other files from my main storage that I'm not working on immediately, they can sit on cloud and/or external drives until later).
   
   
2. Mux the 2003 JP DVD DTS as-is, with zero resync whatsoever, to the 2015 BD and see what happens. Frankly, it looks like (from the waveforms) the sync is actually incredibly close already at the beginning, and it does drift throughout, but for the most part it is remarkably close to the 2015 remix audio (if you just look at the centre channels, I mean). Depending on how that goes, I might try some different delays to see if just shifting it back and forth a tiny bit (which will make some sections better and others worse, in theory).
   
   
3. Compare the 2003 JP DVD DTS against the 2008 FR BD DTS-HD MA once I get it, to see if (and if so, how) it differs in sync. From the info I have so far, it would appear that they should theoretically be frame-identical in terms of video/film frames, so their sync should be extremely close if not the same... unless they're *not* frame-identical, in which case...
   
   
4. Attempt to accurately IVTC the 2003 JP DVD in order to ascertain whether it is indeed frame-matched to the later releases from what seems to be the same master (e.g. 2008 JP & FR BDs, 2015 US BD). This should allow me to definitively tell whether any adjustments are needed for sync with those later releases, with me doing all calculations using nothing but numbers of video frames (converted to audio samples, and in turn, DTS frames, for the best possible accuracy without reencoding or decoding at all). This (without the need for IVTC) is basically how I dealt with Snowpiercer, but that was just dealing with different regional video encodes of the exact same digital video (and yet there actually *were* missing frames! MADNESS!) The quickest but by no means most storage-efficient way of doing this thus far has proven to be demuxing the H.264 and repackaging it as a .mov file, opening it in QuickTime and just browsing through it at specific points to see where certain frames fall within the file. However, with there being different encodes from different generations and media here (DVD as well as BD, not just different encodes of the exact same BD master) it would probably be more sensible to transcode them all to a format that I can look at on a timeline in something like Resolve or the Adobe suite and then I should be able to just skim through the timeline watching for any frames that do not exactly match across several different video tracks. But that's going to need a large amount of free storage on my internal drives, and right now I don't even have close to enough space.
   
   
5. Repeat this process for other T2 audio if necessary for sync, such as the two commentary tracks and the remixed audio, but using whatever the best sources appear to be (assumption would lead me to look toward the UK Skynet Edition for the 6.1 DTS-HD Master Audio since it doesn't necessarily rely on your AVR accurately decoding the matrixed centre surround channel, but I have no clue whether this is sonically better or worse than other alternatives such as the Japanese TrueHD or US Skynet Ed, both of which are 5.1 matrixed encodes of the 6.1).
   
   
6. Assuming I'm able to do a reasonable job of this, package it up as an audio-only deal, since there are plenty of releases out there to source the video from (I'd specifically recommend the 2015 US BD from what I've seen thus far but I have no illusions about it being perfect or anything). I may repeat all of the above for the 2017 restoration as well, for the sake of completeness, but as of yet I have not yet bought that one because from my perspective it stinks of the same spirit of revisionism that I have been bitter to the core of my being about over the Star Wars films since I first realised that they had changed them in the mid-90s. I saw the 3D version in the cinema and I was actually impressed with the actual Stereo D conversion but the amount of DNR and the colour grading of it really put me in a bad mood (plus, obviously, remix audio again). I haven't seen the 2D, 3D or 4K BD releases yet.
   

This is not a million miles away from what I was (and indeed still am) trying to achieve with the whole "Millions of Voices" thing: pick a sync or two, and archive up as many different audio options as possible, in the best possible quality (either lossless if they were lossless to start with or had to be decoded, or without de- or re-encoding if lossy) although in T2's case there are significantly fewer tracks that strike me as important enough to merit archival like this (with SW I've got something like 20-30 tracks per film). I can see me doing this with other films that have similar issues.

[BDgeek]

Yes pipefan, thank for the tip on lddb.com, it's great database I use regularly.

The LD I mentioned was referenced on my post, but not linked, I'll fix that now: 
LD68952-2DD

It's THX certified LD with AC-3 and PCM tracks, released by Live Entertainment in 14/04/1998. That was probably a day and date or very close release with the Live THX DVD we all know about, but with the added bennefit of the PCM track, and maybe, a better AC-3 track, as it was usual for LDs to have better AC-3 tracks than DVDs.

As for the LD you mentioned LD68952-2 , I had the opportunity to compare it to the Live THX DVD in early 2000 on a full THX EX certified HT equipment chain (surround processor/amplifier, speakers (dipolar surrounds) and cables) and while the AC-3 sounded more open with better separation, the LD PCM was much more powerful and robust. All in all, IMO the LD was preferable overall.

Anyway, your project sounds great and I intend to colaborate with the R4 DD 5.1 track and the 2015 track demuxed through DGDemux. It would be great if other members could provide some LD tracks as well.

[Stamper]

You certainly would need the Squeeze LD track, which sounds badass, + the mono mix it has.

[pipefan413]

You certainly would need the Squeeze LD track, which sounds badass, + the mono mix it has.

The later THX one specifically (PILF-2555, 1998) or is the audio the same on both the 1996 (PILF-2187) and 1998 ones anyway?

I had assumed that both were probably comparable or identical to the 1997 US DVD, being 5.1 AC-3 from around the same time. Would be cool to hear the mono and Dolby Stereo matrixed mixes from it in any case, both being pre-2000. Unfortunately though I don't have any LD equipment whatsoever.

[Stamper]

I can only vouch for the first one, as it's not the standard master.

[pipefan413]

Here's something weird I just discovered while syncing the 2008 FR BD to the 2015 US BD (theatrical cut in both cases): it turns out there's a mistake on the 2008 FR BD wherein they use the wrong take of a shot! Either accidentally or intentionally, they've taken the Special Edition take but cropped it, instead of using the theatrical cut take.

The scene is the one with Arnie and Wee John in the car at night, where John's having a go because his big daft buddy model 101 doesn't understand why you shouldn't murder people. In the theatrical cut, the scene ends (around 1h 21m 13s) when Arnie responds to John whinging by doing a slow head-turn to look at him blankly, then it cuts to the incredibly dark home assassination scene starting with the kid's RC car zooming about the floor of the house.

In the Special Edition, the car scene lasts longer, using an alternate take of Arnie's slow head turn that follows it up with a goofy wee half smile. It then cuts back to John's response and then to an establishing shot of the house that isn't present in the theatrical cut either.

What they've done for some reason on the 2008 FR BD is edit the Special Edition take down so that it ends before he does the wee smile. This is obvious mostly because of the scenery in the background outside the driver-side window. The version of the shot that's used in the theatrical cut playlist of the 2015 US BD matches the one on the (theatrical) 2003 JP DVD, so this does appear to be correct, but the 2008 FR BD "theatrical" shot is trimmed from the version used in the 2015 US BD's Special Edition cut playlist.

I've put comparisons on Vimeo but stuck passwords on them, gimme a shout if you're curious to see what I'm on about and I'll link you!

[BDgeek]

Nice finding pipefan413, thanks for reporting it back.

BTW, were you able to check for the differences on the 2015 audio track due to the use of different demuxing tools?

In my experience the results were different using EAC3TO, MakeMKV, MKVToolnix and DGdemux.

[pipefan413]

Nice finding pipefan413, thanks for reporting it back.

BTW, were you able to check for the differences on the 2015 audio track due to the use of different demuxing tools?

In my experience the results were different using EAC3TO, MakeMKV, MKVToolnix and DGdemux.

Not yet, because I wanted to get a full sync done. Aaaaaand I just finished calculating everything to sync the 2008 FR BD to the 2015 US BD! I'll take a look at the two different demuxes of the 2015 remix audio and report back once I've done it. First I'll just do a hex comparison to check if they're bit identical, and if not, I'll take a closer look. But the priority is getting the sync actually done for now. EDIT: bit comparison shows that they're identical up until offset D805ACAA where they suddenly diverge. Don't have time to look into what's different, but what I can do is decode both to PCM and see how far into the bitstream offset D805ACAA actually is (obviously will no longer be offset D805ACAA, but will contain different data as it does in the DTS-HD MA encode) which will then let me look at the waveform and/or spectrogram around that point to see if I can work out if one of the demuxes has a repeated DTS frame (512 samples) or something like that.

There were a whole bunch of frames missing from the 2008 FR BD, which is going to make this more awkward than expected. But what *really* threw me for a loop (apart from that accidentally used SE alternate take I just mentioned in the last post) is something really dumb toward the end of the film...

Y'know that bit where Sarah

comes up to John all injured and pleading, but then *actual* Sarah comes up behind her and reveals that the first Sarah was actually just T-1000 trying to trick John

?

Well, after that, it cuts to John's reaction. On this encode, there are 3 frames missing from that cut: 2 before and 1 after. That's not weird, considering how many frames are missing from this release.

But then I found that there was somehow an EXTRA frame compared to the 2015 a few minutes later. I thought that was weird, because there shouldn't really be any frames on the older encode that aren't on the newer one. So I backtracked until I found it. Turns out, at frame 179759, *they randomly reinserted the missing frame of the John reaction shot from 61 frames earlier*! HOLY CRAP.

[pipefan413]

I'm finding that, perhaps unsurprisingly, sync on the 2008 FR BD is not very accurate in the first place. Most if not all of the time the audio appears to be a bit behind the video (which isn't necessarily a huge problem in playback, the speed sound travels at being slower than light and all that) by various amounts but it looks like quite a lot of the time the difference is about 1 video frame's worth. If I knock 917 ms off the start of the 2003 DVD DTS, it's very close to the 2008 BD DTS-HD MA but not identical, with the '08 starting out behind but then later getting ahead. This has me wondering if maybe they did not fully account for missing video frames in the audio sync of the 2003 JP DVD, which would actually be a good thing since it theoretically means less patching to make it watchable with the 2015 BD (they're already in very close sync at least at the start and do indeed have the same number of black frames before the Carolco logo), possibly even less than I was already doing to approximate 2015 remix sync. Will need to experiment.

It also looks like in the moments where several frames are missing, the audio isn't actually corrected at the time on the 2008 FR BD, though it must be corrected overall because it remains mostly about the same sync if you check different places in the file for audio that happens on specific video frames. This is arguably sensible if it would be noticeable e.g. when John is being a dick to "Todd" and there are 3 frames missing from the video, it does not seem to skip in the audio despite 3 frames amounting to a 6006 sample discrepancy (he's talking with You Could Be Mine blaring from a boombox and the song doesn't seem to jump). Maybe for bits like this they just correct as soon after (or before) as possible and leave the moment itself alone so sync drifts until fixed. Maybe that's also what I'm going to have to do as well; I had already started to use this approach in some places when I was trying to approximate the sync of the 2015 remix track.

[Kreeep]

So, my brother and I have been looking into a bunch of different T2 video transfers and corresponding audio in the past couple weeks. There's definitely some strange discrepancies between different releases that makes syncing up audio without editing very difficult. As you stated, the 2008 FR BD is missing frames, not just in the "She's not my mother, Todd" scene, but throughout. There's even an extra frame from the director's cut that wasn't cut out towards the end. I can share timestamps if anyone else wants to work on it. I've done what I can to sync up the 2008 "CDS" BD DTS-HD audio to other theatrical releases. The result isn't too bad, but it's definitely not perfect. Still working on it.

This makes it even more difficult to sync up to the latest theatrical "remaster", because the remaster seems to be the only release that actually includes an additional 3-4 frames when Arnie rides off on his bike before cutting to Sarah Connor's interview recording at the asylum. This film is a frustrating one Within the next week I'm hoping to be able to share a collaborative effort of what I would like to call as definitive a version of T2 as we can get, with a bunch of high quality audio tracks muxed in.