The 1997 OT Special Edition Trilogy Preservation Standards Thread (* unfinished *) — Page 6

This is certainly some encouraging news.  I just want a great looking SE with the legit DTS audio to boot.

The 5 Star Edition was a great idea when those folks put it out but it fell way short of anything it could have been and now WE here, have a chance to take it to a level far better.

Teamwork WORKS!!!!

:) 

You_Too said:

Actually DJ is the one who is synching them, but we don't even know where all those frame drops are? Obviously some are at the reel changes and since your audio is divided into reels it should be easy to synch them reel by reel, but I didn't know they were missing frames in the middle of reels. It would be very nice if you'd point out where.

I don't know about ANH (yet) but TESB is missing the following, assuming 'Star Wars' appears at frame 711. This is where the frames should be, so the numbers may be off by up to 15 frames by the end:

32645 - General Rieekan "prepare for ground assault" (actually around this point, I have been unable to pinpoint the exact frame as there is very little movement where the frame is missing; somewhere just before the officer he is talking to nods, I duplicated this frame in my synch source).

75305 - last frame of the asteroid hitting Star Destroyer bridge, missing from all SE laserdiscs, and broadcasts, but we have previously discussed in this thread and we believe it should be there, but isn't likely due to them all coming from the same master. Half way through reel 3.

75306 - immediately after the last one, officers conference call with Vader, first frame missing. Half way through reel 3.

93230-93241 - The cave, when Luke chops Vader's head off. Video errors, looks like the DVB signal broke up at this point, or some sort of encoding error. Another source will be needed for this shot. About 2000 frames into reel 4.

151133 - Reel change: Luke entering the tunnel after kicking Vader off the platform, first frame missing.

Reel changes should occur at 30988, 59815, 91134, 122461, and 151133, the reel changes for ANH and ROTJ have been posted earlier in this thread (look at page 3, you'll find them), for your reference. I'll take a look at ANH now and see how many are missing.

Ok, the TB broadcast of ANH is in better shape than TESB, missing only 6 frames (assuming "Star Wars" title card at frame 711):

50193 - close-up on Luke listening after Ben says "...it binds the galaxy together." May not be exactly this frame as its difficult to tell but certainly a frame is missing from this shot. Middle of reel 2.

85223-85225 - last 2 frames of Alderaan exploding, plus first frame of Luke training with his lightsaber on the Falcon. Towards the end of reel 3 (reel 4 begins at 90295).

147671+147672 - last frame of the astromech being lowered into the X-Wing with the mechanic waving his arms, first frame of next shot looking down on an X-Wing. Reel change.

Thats it. I'm surprised there weren't more frames missing around the reel changes, and am even more surprised there were so many missing mid-reel. I wonder if there were ad-breaks in the original broadcast that could be responsible for some of these? I know that the Reivax broadcast is missing 15 sequential frames during the Ben/Vader duel alone!

I hope this is useful. I've been keeping an eye on your release thread at the work you're doing, and downloaded DJ's samples and both are looking exceptional considering your sources! I'm looking forward to checking them out once they're done!

Thank you so much for that info.

Should the Star Wars title appear at 711 in all three films?

And doesn't ROTJ miss any frames? Other than the reel changes.

EDIT: Isn't the logo on the top right of the TB versions the Canal+ logo? If I remember correct, they don't have commercial breaks in movies since it's a pay-channel.

You_Too said:

Thank you so much for that info.

Should the Star Wars title appear at 711 in all three films?

And doesn't ROTJ miss any frames? Other than the reel changes.

I haven't checked ROTJ yet, I'll take a look at it tomorrow.

I think msycamore's figures were 711 for ANH and ESB and 712 for ROTJ, but I've been working with all 3 starting at 711 for uniformity. I guess its preference, but so far the numbers posted up here have worked out nicely; the first DTS synch for ANH that I uploaded a while back was synch'ed perfectly to msycamore's .m2v, which I didn't even have at the time.

Incidentally, the SE frame count for ANH is identical to the '04 DVD. Even with the additional changes each shot is the same length.

You're most likely right about Canal+, just trying to figure out why those frames were missing, but I guess it could be due to any number of reasons. I wonder if the laserdiscs have the same frames missing.

You know what would be great....to have all the numbers in 1 post.  Frame counts are obviously important for synch.  Is there any chance those of you who are good with this angle could accumulate the SE frame counts with reel changes to help some of us better understand the synch of the video/audio of these 3 movies?

Huge thanks to ALL who are making this possible.

Cheers!!!!

Jetrell Fo said:

This is probably why 24fps works best...

    Film uses 24fps, and more recently, 48fps and even higher. Timecode in theatrical films is different and very simple — it counts every frame from number 1 to the total number of frames in the movie.

While this is true for editing, an actual 24fps signal (as opposed to a 23.976fps one) isn't always fully compatible with modern displays.  For example, the Total Recall "Mind-Bending Edition" Blu-ray is one of the few discs encoded at 1080p24, and a few people had problems where frames are skipped/dropped: http://www.avsforum.com/search.php?containingthread[]=1406134&output=posts&action=disp&search=24&resultSortingPreference=recency

If possible, I think the final release should be at 23.976p.

Buster D said:

Jetrell Fo said:

This is probably why 24fps works best...

    Film uses 24fps, and more recently, 48fps and even higher. Timecode in theatrical films is different and very simple — it counts every frame from number 1 to the total number of frames in the movie.

While this is true for editing, an actual 24fps signal (as opposed to a 23.976fps one) isn't always fully compatible with modern displays.  For example, the Total Recall "Mind-Bending Edition" Blu-ray is one of the few discs encoded at 1080p24, and a few people had problems where frames are skipped/dropped: http://www.avsforum.com/search.php?containingthread[]=1406134&output=posts&action=disp&search=24&resultSortingPreference=recency

 

If possible, I think the final release should be at 23.976p.

Very possible and very easy, I assume that's what You_Too and DJ are going to do. For sync the DTS audio needs to be matched at 24fps (because it fits perfectly) but converting it to 23.976 should just be a matter of putting it through eac3to to a lossless format then encoding to DTS.

I don't believe that most modern displays/players would have this problem, the standard has been set and most currently available and future hardware should support 24fps without dropouts.

That said, your point is well taken and you are right. The conversion from 24 to 23.976 is so easy to do and is more compatible for those with older hardware/video processor setups that it would be crazy to leave the final product at 24fps, I guess.

Ok, here are the 'standardised' reel change numbers from the previous posts

ANH (originally posted by Darth Mallwalker):

0 = first frame

711 = STAR WARS title card appear

28343 = reel 2

59025 = reel 3

90295 = reel 4

117270 = reel 5

147672 = reel 6

ESB (originally posted by msycamore):

0 = first frame

711 = STAR WARS title card appear

30988 = reel 2

59815 = reel 3

91134 = reel 4

122461 = reel 5

151133 = reel 6

ROTJ (originally posted by msycamore):

0 = first frame

712 = STAR WARS title card appear

25055 = reel 2

54317 = reel 3

84511 = reel 4

112587 = reel 5

142918 = reel 6

172032 = reel 7

Also, I haven't checked the TB of ROTJ yet, but msycamore has previously reported "11 frames around the reel breaks and a few more at various points" missing.

CapableMetal said:

Ok, here are the 'standardised' reel change numbers from the previous posts

 

ANH (originally posted by Darth Mallwalker):

0 = first frame

711 = STAR WARS title card appear

28343 = reel 2

59025 = reel 3

90295 = reel 4

117270 = reel 5

147672 = reel 6

 

ESB (originally posted by msycamore):

 

0 = first frame

711 = STAR WARS title card appear

30988 = reel 2

59815 = reel 3

91134 = reel 4

122461 = reel 5

151133 = reel 6

 

ROTJ (originally posted by msycamore):

0 = first frame

712 = STAR WARS title card appear

25055 = reel 2

54317 = reel 3

84511 = reel 4

112587 = reel 5

142918 = reel 6

172032 = reel 7

 

 

Also, I haven't checked the TB of ROTJ yet, but msycamore has previously reported "11 frames around the reel breaks and a few more at various points" missing.

Thank you CapapbleMetal, for posting it like this.  Seeing it like this tells me what frame a reel begins and ends at.  I learn better when I have a "frame" of reference (sad joke, I know) to go by.  I appreciate it.

My next learning experience will be to put together the audio for TESB & ROTJ like you did for ANH.  I've got all those mono files in each folder ..... I'm still trying to figure out how to combine them all properly in each folder so I can make 1 big audio file ... PMSL.  Since my only real experience with audio aside from listening has been some Vinyl archiving, I'm getting a real work out moving to surround sound.

Darth Mallwalker said:

Not 15, it's 50! (132114-132163)

Wow thats loads! I did count them, but it was several months ago. Fifteen kind of sounds like fifty. That's my story and I'm sticking to it ;) You are of course right, if I remember correctly (which I haven't!) its a big chunk from them dueling in the corridor just before they are shown in the doorway.

I wonder if the laserdiscs have the same frames missing.

Over the past weeks I've been comparing US & Japan LD captures.
Not my own caps-- I don't own those LDs, but ones available on usenet.
Both sources are raw 29.97fps pulled-down which is good from a certain frame-counter's point of view.
Lots of IVTC'ing ensues and by this point we got sync scripts for US trilogy & Japanese sequels.
US & Japan have different side breaks in some cases, which is also good for frame-counters like us.

Looking back at some notes from Sept, the US ANH LD lacks
85225-85226
147671+147672
and it's missing one field of 162630.

Should be able to provide more specifics for sequels in the coming days.
Earlier in the thread I agreed with msycamore about inserting 75305 into ESB.
After noticing variation between US & Japan versions I've had some radical new idea about that.
If we're to believe the audio evidence, perhaps two frames are missing instead of one.
I need to present that evidence here.

ROTJ frame counting is finished but I haven't done any audio analysis of that one yet, so I can't comment on msycamore's proposed numbers.

 

Excellent. I do own both NTSC and PAL laserdisc sets, just need to check them. TB broadcast of ESB is missing 75305 and 75306. I'll give the NTSC and PAL sets a look tonight.

ROTJ may be correct with the numbers myscamore posted before, I did manage to sync the DTS to that at 24fps. The only piece that in theory should differ from the DVD is the insert of Naboo, it could well be a case that, because its in the last reel, I may be off by the end by a frame or two. Maybe we could figure out the first frame of the last shot to confirm we've got the right length, in addition to confirming the reel change points?

Thank you so much guys for helping out with this.

Good idea about the 2004 SE. I'll probably try to synch the 97 SE completely to that, since it would make DJ's audio synch job a bit easier since we have multiple audio tracks. He could just synch them to the 2004 SE's, with the exception of ROTJ like you say about the Naboo shot, which have to be cut.

50193 - close-up on Luke listening after Ben says "...it binds the galaxy together." May not be exactly this frame as its difficult to tell but certainly a frame is missing from this shot. Middle of reel 2.

I compared it to the GOUT and it looks like the missing frame should have been the very first one of that shot.

Towards the end of reel 3 (reel 4 begins at 90295).

There's an unknown frame missing there? I guess I could pinpoint it once I have it synched to the 2004 SE.

32645 - General Rieekan "prepare for ground assault" (actually around this point, I have been unable to pinpoint the exact frame as there is very little movement where the frame is missing; somewhere just before the officer he is talking to nods, I duplicated this frame in my synch source).

The frame missing should be between 32732-32733. (I have the SW title appearing at 711 now so it should be correct.)

We'll probably use mvflowfps to recreate that frame. (Part of avisynth package mvtools)

I've added the frame count for the movies to the first post in the thread.  As the numbers get sorted and agreed upon I will update them accordingly.

Any suggestions?  Please post them so I can incorporate them in.

msycamore said:

Darth Mallwalker said:

 

CapableMetal said:

the first frame of the last shot

Do you mean the last shot before wipe to credits?
(My 2004 DVD is sealed in polythene, so I can't even count that one.)
Using msycamore's proposed numbers (which I'm not endorsing yet) it came out to 183,826 in my script.
183.825 is the last frame of Mr.Shaw's ghost.

Are we in sync?

 

:) I decided to delete the Jedi files when I saw the quality of those sources, and I wouldn't trust those Jedi numbers I posted either before more investigation is done. I just did a quick compare against the US NTSC DVD, and those shouldn't be regarded as the final word on frame counts as the ESB DVD lack a few frames that are present in the '97 SE DVB's for example.

Those frames mentioned that are absent in the TB ESB can easily be restored by using the G'Kar (except 75305), that's how I did it.

As long as there can be an "as accurate as possible" frame count, it will all help out in the end.

Is there any way to determine frame count from the DTS audio?  Or at least a way it could help confirm them?  I did appeal to someone who might have some access to the 35mm prints but I have not heard anything back as of yet as to whether it was possible to get such counts, "per reel", to make it easier to match via the DTS audio.

You_Too said:

There's an unknown frame missing there? I guess I could pinpoint it once I have it synched to the 2004 SE.

Its not unknown, its 3 frames from near the end of reel 3. 85223 and 85224 are the last two frames Alderaan exploding, and 85225 is the first frame of Luke lightsaber training on-board the Falcon.

Darth Mallwalker said:

Do you mean the last shot before wipe to credits?

(My 2004 DVD is sealed in polythene, so I can't even count that one.)

Using msycamore's proposed numbers (which I'm not endorsing yet) it came out to 183,826 in my script.

183.825 is the last frame of Mr.Shaw's ghost.

 

That's exactly what I meant ;)

My synch source has the same. I think the numbers may be accurate as I made my sources by aligning several different releases side by side and realigning each where there were sections missing. By then end of it the only release not missing frames was the DVD.

The only thing I am unsure of is where frames were missing from each except the DVD. Should they be there or not?

Jetrell Fo said:

Is there any way to determine frame count from the DTS audio?  Or at least a way it could help confirm them?  I did appeal to someone who might have some access to the 35mm prints but I have not heard anything back as of yet as to whether it was possible to get such counts, "per reel", to make it easier to match via the DTS audio.

I'm not sure there is. There is overlap at the start/end of the audio in the reels and I tried the "sample-sample" trimming method, but the waveforms for the same parts look different on different reels, likely due to them being recorded in different takes from an analogue source to digital (at the end of one of the ANH reels the audio slows and drops in pitch, like a tape being stopped slowly). This makes it difficult to get them aligned mathematically perfect, which is why there is no exact science to synchronising these DTS reels other than putting them in the place they should be within a video source and making small adjustments until its perfect.

http://forums.afterdawn.com/t.cfm/f-37/editing_trimming_a_dts_wav-786182/

This program might be of help.

CapableMetal said:

You_Too said:

There's an unknown frame missing there? I guess I could pinpoint it once I have it synched to the 2004 SE.

Its not unknown, its 3 frames from near the end of reel 3. 85223 and 85224 are the last two frames Alderaan exploding, and 85225 is the first frame of Luke lightsaber training on-board the Falcon.

Sorry, I got it wrong. I thought you meant that after those 3 missing frames there was another one missing at the end of reel 3.

Darth Mallwalker said:

If target is true 24fps then it's only one step.
Use your favorite software (iZotope RX Advanced if you can afford it)
and convert the DTS files from 44100 to 48000Hz.

Now if the target will be 23.976 it could be a two-step process.
First 44100 to 48000, then 24 to 23.976
But that's resampling twice, and we'd like to avoid that if possible.
We can avoid it! Here's how.
Use iZotope or whatever to convert from 44100 to 48048Hz. Not a typo, I did say 48048.
Then using (SoundForge, Audacity, Audition, Vegas?) to edit the WAV's properties and force the rate to 48000, without resampling.

It's a bit like using AssumeFPS() in the video realm-- it doesn't change the total frame count.
When you force 48048 to 48000 in the WAV's header you're not changing the total number of samples, but rather slowing them down to NTSC speed.
Slowing down without costing the extra step. Does it make sense?

 

Absolutely right. That's how I've done mine, or I should say doing. I haven't done a SRC to my synchs yet. ESB is doing it now.

Just for the record, you can do it by going to 44144 and SRC to 48000 but its technically less accurate (with a remainder of .144 instead of .048). That's how I did my early synch of ANH that was available on MySpleen, which I've now surpassed.

I have since synchronised the whole trilogy at the native 24fps, 16-bit, 44100KHz (with volume reduction to make it more friendly with DVD/BD's) and am going down this route to match it to 23.976. In fact I've got ESB going through its SRC right now as I finished colour correcting my video source.

In a couple of hours when its ready to mux I'll give it a test and confirm that it works perfectly this way, which I'm entirely confident it will.

Thanks Darth Mallwalker for that explanation.

And while it's lossy in theory then, it did sound pretty damn good and accurate after eac3to did both those conversions. Though it sure looked like it was doing them in one step. Maybe eac3to is that smart? haha

eac3to is an excellent program, especially good at converting PAL audio to/from NTSC with no noticeable quality drop.

At the end of the day the best tools available are your ears and your judgment. If you hear no difference and you're satisfied then its probably good enough.

That said, I can confirm that Darth Mallwalker's method works perfectly and maintains synchronisation with the video at 23.976fps (no surprise there, of course!). I'm currently watching ESB having resampled the audio and adjusted the sample playback rate and everything is in sync and probably as pure as its going to get, including the volume adjustment; technically it could have been avoided but the source reels really are just too loud at their standard levels.

CapableMetal said:

eac3to is an excellent program, especially good at converting PAL audio to/from NTSC with no noticeable quality drop.

At the end of the day the best tools available are your ears and your judgment. If you hear no difference and you're satisfied then its probably good enough.

That said, I can confirm that Darth Mallwalker's method works perfectly and maintains synchronisation with the video at 23.976fps (no surprise there, of course!). I'm currently watching ESB having resampled the audio and adjusted the sample playback rate and everything is in sync and probably as pure as its going to get, including the volume adjustment; technically it could have been avoided but the source reels really are just too loud at their standard levels.

The standard levels being too loud...are you speaking about the way DTS has them set via the whitepaper on it?  My other question I guess is going to be, are you going to release a set based on your work?

:)