[Proposal] Halloween 2020 pipefan413 preservation: THE EXORCIST (1973)

[BDgeek]

Congratulations on the Work pipefan!

Great collection of rare LDs. My vote is definitely for Halloween and Near Dark!

[pipefan413]

I'm wondering if this might be the trouble:

This is done using the Media Express software and software version 10.11.4 (Windows), a combination that produces perfect sync. (Later versions have issues with sync drift, loss of video sync, or both.)

https://forum.lddb.com/viewtopic.php?p=1...84#p109260

I'm using a Decklink Studio (very similar card but with more analogue connections) and the Blackmagic software version 11.6. Maybe the driver is the problem. About to try rolling back to 10.11.4 to see if that makes any discernible difference.

It did!



So reverting to the older version and recording directly with Media Express rather than AmarecTV appears to achieve accurate sync. It also produces 24-bit audio instead of 16-bit, which is odd because I thought the card was incapable of that, but there you go, that's good too. Less good: it produced GIANT files. I'm not sure if these are completely uncompressed or just using a Blackmagic codec that's extremely inefficient compared to things like UtVideo, MagicYUV and Lagarith, but this could kinda be a problem for long-term storage; I guess I could just pipe every raw capture through AviSynth and encode to one of those three more efficient codecs, but that's a lot of processing time added into the workflow. For now, though, I guess it'll do, because AmaRecTV is stuck in 16-bit as far as I can tell and the 24-bit capture is a big plus for analogue audio editing. But yeah, if it works, it works!

[spoRv]

I vote for Braindead too - just put the laserdisc video "as a bonus"!

[pipefan413]

I vote for Braindead too - just put the laserdisc video "as a bonus"!

Apparently the video has some crazy s**t happening like a bit in black and white for no apparent reason but I've not watched it yet. Definitely capturing video for all of these, not just the audio.

Also eh... BRAINDEAD isn't part of the deal, I'll do that after Halloween hahah! But do vote in the poll at the top for whatever your *second* favourite would be... ;-)

[pipefan413]

Right, after a bit of faff (sync was still off on one of my captures) I now have a complete capture of the second pressing of THE EXORCIST from my single-sided player, which appears at first glance to have slightly superior analogue audio performance to my auto-side-changing player. For one, the level of the right channel on the newer player appears to be slightly higher than the left, which obviously isn't great, although it's easily fixable in the edit by bringing the gain up to match, but I digress.

I verified sync as follows:

Code:

SIDE1 = AVISource("CLD1750_NJEL-01007_side1.avi")#.Trim(0, 95071)
SIDE2 = AVISource("CLD1750_NJEL-01007_side2.avi")#.Trim(147, 95993)
SIDE3 = AVISource("CLD1750_NJEL-01007_side3.avi")#.Trim(100, 28186)

SIDE2_16 = RaWavSource("audio\CLD-1750_NJEL-01007_side2_16bit.wav")
JOINED = SIDE1 + SIDE2 + SIDE3
#TEMPSOUND = WAVSource("audio\NJEL-01007_16bit.wav")
#return AudioDub(JOINED, TEMPSOUND).ShowSMPTE().ShowFrameNumber(x=20, y=34)
return AudioDub(SIDE2, SIDE2_16)

This is basically a snapshot of a script I edited a few times to pipe stuff to VirtualDub to check it wasn't doing anything that wasn't precise enough for my purposes. I trimmed the dead frames off the ends of the 3 captures, joined them into 1 file, then exported the joined audio. I converted that in eac3to to 16-bit in order for VirtualDub's audio waveform display thing to work (it doesn't like 24-bit); I could've done this 24-16 conversion in VirtualDub but I was going to potentially want the 24-bit file later so I just made a 16-bit version off that instead. I fed that 16-bit audio back into the script and looked it it in VirtualDub to make sure it lined up exactly. Finally, in case that trimming and joining was cutting anything off that I might need (e.g. if sync was off and some of the actual audio was being cut off by the trim and join), I repeated that process without the trimming part and exported audio for each raw capture separately, then looked at the same frames in VirtualDub. They were exactly the same, so I'm happy that this has perfect sync. Here's the start of side 2 for reference:



When it's trimmed and joined back to the other 2 sides, there is no difference in how the audio lines up, so I'm happy that it's not cutting off anything apart from background analogue "silence".

There's one slight issue with using Blackmagic's software to do the captures: it appears to be saving it as uncompressed, raw video data, whereas AmaRecTV was using whatever I told it to use to compress e.g. Lagarith/UtVideo/MagicYUV. Part of the trouble with this, apart from the obvious massive file size, is that it would appear that I can only get AviSynth to output the colour as RGB32 for some reason, despite the Blackmagic Media Express software indicating that it's actually "8-bit YUV" (which according to MediaInfo is specifically UYVY. I can tell AviSynth to use the UYVY decompressor, but it still outputs RGB32 instead of leaving it as 4:2:2 YUV, and if I try to specify that the colour is to be kept as YUV then it doesn't display correctly so I must be mismatching the actual colour space with what I think it is. This may be a complete non-issue if I select the appropriate YUV colour space when compressing using the following script:

Code:

SIDE1 = AVISource("CLD1750_NJEL-01007_side1.avi", pixel_type="FULL").Trim(0, 95071)
SIDE2 = AVISource("CLD1750_NJEL-01007_side2.avi", pixel_type="FULL").Trim(147, 95993)
SIDE3 = AVISource("CLD1750_NJEL-01007_side3.avi", pixel_type="FULL").Trim(100, 28186)
JOINED = SIDE1 + SIDE2 + SIDE3
return JOINED

So, if I feed that to VirtualDub2 and then save the video as Lagarith with the appropriate pixel format (YUYV/YUY2), it's effectively converting from UYVY to RGB32 then back to YUYV, despite YUYV and UYVY being pretty much comparable. I dunno if that conversion is a problem or not. For this capture it's irrelevant because I'm not really using the video apart from as a reference, but for later captures I want to know that I'm not screwing with the colour for videos I want to actually keep for archival and viewing purposes.

For comparison's sake, the captures from AmaRecTV that I was saving as Lagarith were YUY2/YUYV in the first place and AviSynth output YV16 instead of RGB32. I was converting to YV12 deliberately at that point because I wasn't fussed about the video quality anyway for what I was doing at the time.

Anybody who knows more about these things able to shed some light?

EDIT: Here's one idea. I realised I can use VirtualDub2 in 64-bit mode to quickly convert the uncompressed UYVY to Lagarith *without* converting it to YUY2, although it decodes as YUY2 anyway (takes less than 10 mins per LD side). Still, that's possibly better than a conversion to RGB32 and back again (turns out, yep, it definitely is because it's just a byte order swap). Plus, means I can delete the giant raw captures without having to only save them in joined form for archival, since I can instead keep the Lagarith parts as separates and then my AviSynth script will be what joins them. Although I may keep a small proxy after IVTC, I've not really decided yet. I also tried RawSourcePlus, although it would seem I'd have to work out how big the header is to know where to start the index.

As you can tell, much of this is still me figuring out a workflow that I like. Once I get into the swing of things I think I'm going to love this, but I'm still deciding how I like to work.

[pipefan413]

Jeeeeeeesus... finished figuring out the structure of THE EXORCIST resync and I kinda want to dig a big hole and jump into it because, to quote my own AviSynth notes...

# 34 + 35 = 69 frames missing at 1st reel change
# 33 + 28 = 61 frames missing at 2nd reel change
# 1 + 8 = 9 frames missing at 1st LD side change
# 12 + 9 = 21 frames missing at 3rd reel change
# 25 + 19 = 44 frames missing at 4th reel change
# 31 + 11 = 42 frames missing at 5th reel change
# 10 frames missing at 2nd (final) LD side change
# 10 + 9 = 19 frames missing at 6th (final) reel change

# Total missing frames:
# 69 + 61 + 9 + 21 + 44 + 42 + 10 + 19 = 275
# 275 / (24000/1001) = 11.470 seconds (!)
# x 48000 = 550,550 samples at 48 kHz

No wonder @The Aluminum Falcon was struggling to make it work well! I dunno how the hell he managed what he did, looking at the sizes of the holes in the LD source. Bloody hell.

We're really talking about gaps that are way too large to bridge with extension crossfades, I think, and there is no alternative source that I know of apart from my other LD copy (which is the first pressing of the same disc, so almost certainly has identical frames). If I were to attempt to patch this, I'd have to create a custom mono downmix from something else, like one of the 5.1 mixes or perhaps preferably a 2.0 track from something else like one of the newer LDs. But even that is kind of a potential s**tshow; it may not be a transparent transition regardless of what I do with it.

The final alternative would of course be to resync in the opposite direction: remove the frames from the Blu-ray that do not appear on the LaserDisc(s) in order to avoid editing the soundtrack. Normally I would be completely against that, but it may make slightly more sense in this case given that my primary motivation for doing this is "theatrical" reconstruction (I'd be using it with @The Aluminum Falcon's custom regrade/edit of the Blu-ray rather than the Blu-ray itself, since the Blu-ray as it is remains a fairly abhorrent thing from my perspective). I'm honestly tempted to use the information I've gleaned from this analysis to just do the video edit instead, but I'd like to be able to do something to get the audio workable with the Blu-ray as well. If anybody has any clever ideas, I'm all ears!

Incidentally, this also suggested something else... I think that the scenes in the bedroom actually *are* meant to be fairly blue, though maybe not quite to the extent of the Blu-ray. The LaserDisc at least has a very icy off-white tone to it, which differs considerably from the TERROR IN THE AISLES colours used as the reference for TAF's regrade:



The LD may not necessarily represent the original theatrical grading, but it is clearly scanned from a print because it's got every reel change marker intact. But TERROR IN THE AISLES, though also scanned from a print, may not necessarily represent the theatrical colours either. It was made in 1981 I think (or just before) when the film had been released at least twice, so the print may have been from the recent 70 mm re-release, or it may have just been like 7-8 years old and beginning to show signs of age. It certainly does show some significant print damage, so that wouldn't necessarily surprise me. By contrast, the LD print was perhaps more likely to have been struck fresh off a new IP or something for the home video transfer, rather than using an aged 70s release print as TERROR IN THE AISLES looks like it might've. In any case, trying to warm up the colour temperature of these scenes on the Blu-ray transfer has a few side effects, as of course it would unless you get fancy with power windows and the like. For instance, it turns Regan's nightgown from blue to green...



... and the sink is now a bit yellow/orange instead of white:



Very interesting indeed. This thing just keeps getting more and more complicated in ways I don't have good fixes for...

[spoRv]

Very interesting indeed.

And it's also open matte!

[pipefan413]

Very interesting indeed.

And it's also open matte!

A little bit, though quite a lot is chopped off the sides as well!

[pipefan413]

The trouble here is that there is actual dialogue and other conspicuous audio lost in these big holes at the reel / side changes. For instance, here is the first reel change:

pw = Pazuzu

If you look closely, you can see Regan's mouth moving before we hear her speak; on the OOP DVDs and other non-mono transfers, she says "Here it comes..." before she then says "... there!", but those frames are missing from the mono transfer. So I can't just leave it silent, it needs a patch over. The other reel changes are generally similar; there's another one at the moment where the bed is chucking her about ("Mother! Make it stop!") and Chris jumps on the bed to try to weigh it down or whatever, which again is a very noisy moment where they're both yelling and screaming and the bed is rattling so I can't just stretch it out or leave it silent or whatever.

So I'm on the hunt for alternative audio sources to patch over the massive holes in the mono track. What I'm thinking is that if I can find another LaserDisc with an older audio source than the modern transfers but hopefully without the missing frames (or at least without *some* of the missing frames), I can downmix the Dolby Surround track to 1.0, duplicate it to dual mono 2.0, and then very carefully patch the gaps as seamlessly as possible. Whilst I will likely record this digitally as well for posterity, I will most likely do the recording in analogue to slightly better match it to the mono, though it won't be as manky as an actual mono audio track as it'll be turning the digital track to analogue output rather than playing back an analogue track which has no noise reduction on it (which is what the mono track is, being pre-CX). I may even take it a step further and take a leaf out of Belbucus' book: record the digital audio, then record it to tape (either audio casette or VHS) and back again, to intentionally introduce tape hiss to better match the actual mono mix.

In the meantime though, I already have the 1997 US DVD, the one with the 4:3 transfer on one side and 16:9 on the other. What I think happened with that disc is as follows. The fullscreen transfer is quite likely from the same master (or a similar one) as these early 4:3 LaserDiscs, but since they decided for whatever reason that the mono track was unacceptable, this instead has a 5.1 track likely sourced from the 6-track theatrical mix used for the 70 mm screenings back when the film was re-released in 1979, whilst the 16:9 side is most likely sourced from the same master previously used to make the *later* LaserDisc transfers, which featured a Dolby Surround track that I'm guessing is probably from 1979 re-release 35 mm (rather than 70 mm) prints that couldn't hold the 70 mm 6-track sound but could holy Dolby Stereo just fine. So I think I'll at least attempt to patch using this first to experiment with how it sounds and see how feasible it might be. As I say, I expect I'll need to do more than simply patch it in, because the abrupt lack of tape hiss (sometimes for several seconds at a time) will be extremely noticeable. If it does seem doable then I'll try to source a late LaserDisc to grab its Dolby Surround PCM track and then figure out how I'm going to go about recording it to analogue and back again to better match the rest of the mono track.

In other words: this has turned into a significantly bigger job than I expected, which *might* mean I can't get it done before Halloween. But the poll's still open anyway, so I guess Halloween or one of the other options might edge it out regardless!

[pipefan413]

Juuuuuust in case...