[Request] NEAR DARK NTSC audio

[Stamper]

I suspect they might also have used noise reduction on the audio hence the thinny quality.

What happens is SC usually did remaster, back 20 years ago, films destined to Pal DVD.

They were all 1080i, all made from either the IP or negatives, sometimes internegs.

So when they did prepare the masters, the audio was made in another room, different from the color timing room.

In most cases, the guy in charge of audio just speeded it up to Pal including pitch change.

At some point, they discovered they could adjust pitch back while still speeding it up for Pal.

They did it on numerous release, but somehow, no one kept a file with which release if pitched up, and which is not, and at some point they do it on all new releases.

Then Blu-ray and HD-DVD comes by.

The begin by just adjusting the audio Pal masters back to 23.976 or 24, pitch not included, for releases like the Rambo series, as they think every master they have on file is already at the right pitch.

People notice and freak out. They decide they might have done something wrong, and from then on, adjust both speed and audio on all releases.

People notice and freak out again because now some of the releases are pitched to low.

They keep doing this hit or miss mess, as they never go back to the audio magnetic sources. They always use their digital DAT masters on file. Also why it sounds crappy and metallic, DAT was a bad format.

Note that all those transfers date back from up to 20 years. They've been making HD transfers since around 1999.

[pipefan413]

Forgot to mention that when I rented the US Anchor Bay THX DVD from a library a few years ago, I noticed the 2.0 and 5.1 DTS tracks sounded super low pitched (as if the pitch corrected 25fps master used for the UK DVD was sped down to 24fps without keeping the pitch) right down from the StudioCanal logo. I also noticed they reused that same 2000s StudioCanal master for the Lionsgate BD and the UK one, but the audio is strangely different there like pipefan413 had mentioned.

What you described in the first reply is essentially what I feared but I'm aghast to have discovered this happening...

Re. the US DVD, if that has indeed been slowed down from the PAL master then I'll know by comparing the music. If it's flat from a C on the start of the end title theme then it's been slowed from PAL without pitch retention. Fingers crossed, but we'll see. In the meantime, I've lopped off the 2 different StudioCanal logos and repitched the 2.0 tracks from the UK (LPCM) and French (DTS-HD MA) BDs, which sound pretty bad to me presumably because they've been repitched up then back down again in addition to whatever other horrors they may have been subjected to to reduce noise and so forth (Halloween's 2014 TrueHD mono sounds like they lopped off the entire low end trying to get rid of tape hiss).

Thanks all!

Still hunting for (ideally the digital) LaserDisc audio so if anyone is in a position to share that please PM.

[pipefan413]

I suspect they might also have used noise reduction on the audio hence the thinny quality.

What happens is SC usually did remaster, back 20 years ago, films destined to Pal DVD.

They were all 1080i, all made from either the IP or negatives, sometimes internegs.

So when they did prepare the masters, the audio was made in another room, different from the color timing room.

In most cases, the guy in charge of audio just speeded it up to Pal including pitch change.

At some point, they discovered they could adjust pitch back while still speeding it up for Pal.

They did it on numerous release, but somehow, no one kept a file with which release if pitched up, and which is not, and at some point they do it on all new releases.

Then Blu-ray and HD-DVD comes by.

The begin by just adjusting the audio Pal masters back to 23.976 or 24, pitch not included, for releases like the Rambo series, as they think every master they have on file is already at the right pitch.

People notice and freak out. They decide they might have done something wrong, and from then on, adjust both speed and audio on all releases.

People notice and freak out again because now some of the releases are pitched to low.

They keep doing this hit or miss mess, as they never go back to the audio magnetic sources. They always use their digital DAT masters on file. Also why it sounds crappy and metallic, DAT was a bad format.

Note that all those transfers date back from up to 20 years. They've been making HD transfers since around 1999.

Hahah, you posted this while I was typing again! I agree about the noise reduction, I've seen it before with Halloween like I mentioned there (and with others but Halloween was especially terrible and sounds very very similar to the BD audio for Near Dark). I'm crossing my fingers that I might be able to use the AC3 from the NTSC DVD but maybe not if SpaceBlackKnight is correct about that being *too* low in pitch. I'm still looking to source both that and the US BD, anyway. There is also one more release I don't yet have, and it's extortionate to import: the Japanese one, which quite possibly just uses the same masters and thus the same incorrectly pitched audio as the French/German BDs, so might not be worth importing at all.

[Stamper]

As far as I remember, the NTSC is either with the two tracks at correct pitch, or the 2.0 is bad pitch again.

I have it on the shelf, will check it out.

[pipefan413]

As far as I remember, the NTSC is either with the two tracks at correct pitch, or the 2.0 is bad pitch again.

I have it on the shelf, will check it out.

Hopefully it's correct, the PAL one certainly is but obviously 25 fps.

New discovery btw: despite the UK BD 2.0 LPCM reporting as 24-bit, it actually appears to be 16-bit. Trying to import into Audacity only works if I tell Audacity it's 16-bit, unlike the decoded DTS-HD I got from the French BD which imported as 24-bit. The files are indeed different sizes, with the French one being about as much bigger as you'd expect if looking at the same audio at 24-bit vs 16-bit. EDIT: Confirmed this with eac3to log file:

3: RAW/PCM, English, 2.0 channels, 24 bits, 48kHz
[a03] Extracting audio track number 3...
[a03] Reading RAW/PCM...
[a03] Swapping endian...
[a03] Writing WAV...
[a03] Creating file "00000 - 3 - PCM, English, 2.0 channels, 24 bits, 48kHz.wav"...
[a03] The original audio track has a constant bit depth of 16 bits.
[a03] Superfluous zero bytes detected, will be stripped in 2nd pass.
[a03] Starting 2nd pass...
[a03] Reading WAV...
[a03] Stripping zero bytes...
[a03] Writing WAV...

I feel like this gets worse the closer I look...

Anyway, to-do list for this one currently:

1. Repitch & sync French BD 2.0 DTS-HD MA (decoded) - DONE
2. Repitch & sync UK BD 2.0 LPCM - IN PROGRESS
3. Obtain & check US BD 2.0 LPCM & commentary; if commentary is correctly pitched, sync, otherwise potentially repitch (and/or repitch commentary from NTSC DVD / German BD / Japanese BD)
4. Obtain & check NTSC DVD (2.0 + 4.0 commentary)
5. Attempt to obtain & check Japanese BD without crazy high cost

[Stamper]

What version is best PQ wise? I assume the german as the edge as it's a BD-50?

[pipefan413]

What version is best PQ wise? I assume the german as the edge as it's a BD-50?

The German encode appears to be identical to the French one a few months before it. They've just lifted the French release, bonus content and all (despite it mostly being French-specific) then added more bonus stuff from the Japanese release and some more German stuff as well (2 German commentaries and a German visual essay / interview that switches to English to speak to Lance Henriksen). I'm going to compare the picture of those against the Spanish and UK/US in more depth, but provisionally:

- The UK/US ones are fairly low bitrate (21.974 Mbps) and significantly darker

- The Spanish one has a marginally higher bitrate (24.901 Mbps), and I haven't gone through it in detail yet but it appears slightly different from the others... not expecting much though based on my experience of other Spanish releases

- The French and German ones are high bitrate (31.904 Mbps) and also brightened up with some colour tweaks, but that may or may not be a good thing for this film (same scan anyway so I don't expect any more clarity in the blacks, really)

[pipefan413]

What version is best PQ wise? I assume the german as the edge as it's a BD-50?

OK, see what you think: http://www.framecompare.com/image-compar...n/7DKWWNNX (UPDATED: added Spanish frame)

Same frame from 2003 UK (PAL) DVD, 2009 UK BD, 2013 Spanish BD, and 2018 French BD (which is identical to German BD encode). Significant differences in all 3. Seems to be a noticeable green cast on the original 2009 encode, which has been curbed somewhat on the 2013 (Spain) and 2018 (France/Germany) one. All of these BDs appear to suffer from significantly greyer blacks than the old DVD, with the 2018 encode being brighter than the 2009 one and the Spanish 2013 one being significantly brighter still (resulting in incredibly grey "blacks").



To be fair, I don't expect there was much depth of black in the source here, but it's still not great (particularly considering, as the title does indeed suggest, the majority of the film takes place in the dark).

[Stamper]

Looks like the latest german/french encode is the most accurate to the master to me.

[pipefan413]

Looks like the latest german/french encode is the most accurate to the master to me.

Yup. It's still very grey and there's little to be done about the overall quality until there's a new master drawn from a decent source, which I could die waiting for. I'm going to pitch-correct every available 2.0 track and try to compare them and decide which to keep. The 5.1 off the US/UK discs appears to be the best shout for the (English) surround mix because the French and German discs get that one wrong as well and it's already got the correct pitch.

I reckon I'm going to crop off the 2018 StudioCanal logo music, pitch correct each track, and stick the (unaltered) StudioCanal jingle back on the start. I could also just cut off the StudioCanal logo and leave it off, which I'm considering, but it should theoretically be easy enough to stick it back and and thus avoid having to trim the video (which I can only do at keyframes without re-encoding, and I absolutely will not re-encode this).

Candidates for main 2.0 stereo audio (all will need re-sync except French track, if used):

1. Unaltered 2002 US DVD Dolby Digital / AC3 (assuming pitch is correct as-is; lossy, but shouldn't need EQd etc.)
   
2. Pitch-adjusted 2013 French BD DTS-HD Master Audio, decoded to PCM (24-bit; quite tinny, probably due to noise reduction)
   
3. Pitch-adjusted 2009 UK BD PCM (supposedly 24-bit, but seems to actually be 16-bit; also has tinny quality from NR)
   
4. Pitch-adjusted 2009 US BD PCM (supposedly 24-bit, but seems to actually be 16-bit; has tinny quality from NR)
   
5. Stretched 2003 UK DVD Dolby Digital / AC3 (pitch is already correct and does not have horrid tinny EQ of BD tracks but timing sped up to 25 fps for PAL system)
   

Candidates for commentary track (need re-sync):

1. Unaltered 2002 US DVD 4.0 Dolby Digital / AC3 (assuming pitch is correct as-is)
   
2. Unaltered 2009 US BD 2.0 Dolby Digital / AC3 (already has correct pitch, unlike German BD)
   
3. Pitch-corrected 2019 German BD Dolby Digital / AC3 (doesn't need resync but sync is less invasive than pitch correction so almost certainly won't use this)
   

To repitch the incorrectly raised track from the French BD while retaining sync, I did this:

1. Decode DTS-HD MA to PCM (or RAW) using eac3to and import into Audacity.

NOTE: If using eac3to for this, choosing file extension .pcm will result in a big endian byte order (as used by Blu-ray Disc spec), whereas .raw is little endian (more common in Windows, for audio files anyway). It doesn't really matter which you pick, but you need to select the same one when importing the file into Audacity because it'll otherwise result in blaring static caused by reading each byte in the file backwards.

2. Select everything except the StudioCanal logo music by converting the number of frames it occupies to sec and ms and selecting everything from that mark to the end of the file.

The 2018 StudioCanal logo runs from the start of the 1st frame to the end of 516th frame, so 517 frames in duration; to convert to time instead, that's 517 / (24000/1001) = 21.563... seconds.

3. Use "Change Pitch" to adjust down (negative number) by same % as assumed original adjustment to reach PAL speedup from 23.976 fps or from film framerate of 24 fps, testing both for accuracy against Tangerine Dream soundtrack CD. I did both because I don't know without working it out whether the master they originally started with before the PAL conversion that caused the pitch shift was running at 24 fps (film rate) or 24000/1001 fps (NTSC). Note that I'm not adjusting speed here but only adjusting pitch by the same factor that the speed must have originally been increased in order to reach the incorrect pitch.

Calculations for this as follows...

If 23.976 -> 25 -> 23.976 fps: -(25 - 24000/1001) / 25 = -4.096% correction needed
If 24 -> 25 -> 23.976 fps: -1/25 = -4% correction needed

4. Export as 24-bit RAW PCM. This results in a little endian file, so I then switched it back to big endian for Blu-ray Disc compliance (eac3to audacityoutputfile.raw finalpcmfile.pcm).

5. Mux verified correct version back in with the AVC video track to regain original sync