logo Sign In

The GOUT Sync Thread — Page 2

Author
Time
 (Edited)

Maybe for the version we have released for ESB, but we have multiple prints of it, along with those for Star Wars and Return of the Jedi.

Team Negative1

Author
Time

Not sure if this is possible, or to what level it could be applied but - is there any chance this could be used to merge the Chroma from Harmy's Despecialized with the Luma from the Renegade Grindhouse?

Preferred Saga:
1/2: Hal9000
3: L8wrtr
4/5: Adywan
6-9: Hal9000

Author
Time
 (Edited)

nightstalkerpoet said:

Not sure if this is possible, or to what level it could be applied but - is there any chance this could be used to merge the Chroma from Harmy's Despecialized with the Luma from the Renegade Grindhouse?

You'd need an excellent image registration algorithm, at least.

Even the best ones screw up constantly, so you'd have to manually check every frame for alignment issues.

It's usually more trouble than it's worth.

poita and I looked into this a few years ago. The best image registration program he could get took four minutes per frame on his absurdly fast hardware and it still wasn't really as trustworthy as it should have been.

A picture is worth a thousand words. Post 102 is worth more.

I’m late to the party, but I think this is the best song. Enjoy!

—Teams Jetrell Fo 1, Jetrell Fo 2, and Jetrell Fo 3

Author
Time

Yeah, I do this all the time for single shots and it's a big ol' pain in the ass, so doing it for a whole movie would be more trouble than it's worth.

Author
Time

Danfun128 said:

Or any other average 35mm release, apparently.

 As I understand it, projectionists would routinely splice the reels together onto a big platter and then cut them apart again to return them to the reels. Obviously to run the film without heads and tails appearing on screen at each reel change they had to be spliced where the real picture information is. Over time, the frames at each end of the reel would get damaged and be cut off. Since the film soundtrack is on the film too, this would not cause audio sync issues and really nobody would notice one or two missing frames. From what I have seen of Team Negative One's prints they are nearly all missing some frames at the beginning and end of each reel when compared to the GOUT.

TheStarWarsTrilogy.com.
The007Dossier.com.
Donations always welcome: Paypal | Bitcoin: bc1qzr9ejyfpzm9ea2dglfegxzt59tys3uwmj26ytj

Author
Time

Silly question, but do Puggo's releases have some of these missing frames, or would a completionist have to upscale frames from the JSC or GOUT where appropriate?

Nobody sang The Bunny Song in years…

Author
Time

All early laserdisc releases, including the JSC, are missing many frames at reel changes.

It's enough that you can actually hear some pretty big jump-cuts in the soundtrack, where portions of the audio are simply gone.

The Definitive Collection (and consequently the GOUT, which was derived from it) is the first home video version where this doesn't really happen.

Author
Time

So, out of curiosity, what release has the most intact frames? The Def Collection? Faces? Gout? Not counting the changes, do the 97se laserdiscs have frames where the other releases don't?

Nobody sang The Bunny Song in years…

Author
Time
 (Edited)

The IP/IN that was used for the THX LD's (GOUT) is a copy of the original negative and IP's and IN's are used in the process to make theatrical prints, so obviously it doesn't include more frames than what is normal. Hopefully it includes the fully edited film. What happens when theatrical prints are starting to lose frames at the reel changes over a theatrical run is another thing. 

That said on the first film the NTSC Bonus Feature is missing about one single frame, so at this time I think it's quite unnecessary moving the goalposts. Such an edit can also easily be done by yourself if the missing frames are restored to whatever project.

hairy_hen said:

I'm actually fairly certain the GOUT does have more frames than might normally have been seen. The reason I say this is that if you listen extremely closely, you can actually hear small jump-cuts in the soundtracks where the audio has been looped, in order to extend it in length. Such a thing would only have been done if the video ended up being slightly longer than the audio, for the sake of maintaining synch.

Without fail, edits of this type occur each time there is a reel change. Since it happens about every ten to eleven minutes, these would correspond to the shorter reel lengths of a negative or interpositive, rather than the double length of a theatrical print. These are the same spots where different video transfers go out of synch with each other.

Most of this had to have been done for the Definitive Collection laserdiscs themselves back in 1993, but the GOUT also has a few additional edits of this nature that the laserdisc tracks do not, though for what reason I'm not sure. Since there can be so much discrepancy in frame counts between versions, even ones derived from the same master, picking one convenient reference and sticking to it—namely, the NTSC version of the GOUT—is the best way to ensure that audio synch issues are eliminated. I don't especially like the idea of dropping any frames either, but in practice the differences are small enough not to be noticed when watching, and it is still more complete than a typical 35mm print (ie, the -1 version) would have been.

The audio is about the last step in the chain, IP's or IN's do not come with a soundtrack. That the audio tracks have small jump-cuts and have the audio looped at certain points to match the video master could be for many various reasons, a few audible edits might even be present on the master tapes and on the print master.

Danfun128 said:

So, out of curiosity, what release has the most intact frames? The Def Collection? Faces? Gout?

I believe it's Faces. Technidisc SWE comes close but IIRC it's missing a field at a side break.

We want you to be aware that we have no plans—now or in the future—to restore the earlier versions. 

Sincerely, Lynne Hale publicity@lucasfilm.com

Author
Time

what makes the Faces have more frames than the gout, and what about the other films in the trilogy?

Nobody sang The Bunny Song in years…

Author
Time

So... what AviSynth script would you use to convert the frames from the PAL DVDs to 23.976 FPS, so that it's in progressive film-speed frames and would (more-or-less) sync to the NTSC audio?

I'm starting to get the hang of DGDecode/MPEG2Source and TFM/TDecimate for NTSC discs, but getting it working for PAL eludes me.

Project Threepio (Star Wars OOT subtitles)

Author
Time

AssumeFPS(clip,"ntsc_film",sync_audio=true)

this will change the fps from 25 to 23.976 and sync the audio with the new frame rate. I did this with 9 seasons of seinfeld and it works pretty decent. It sounds like you want to use the ntsc audio though so you could just add killaudio() after this and then use audiodub to mix the ntsc audio with the pal footage. Of course it would be better to encode the video and then mux in the audio later so as to avoid re-encoding but anyway here's how to do it in avisynth:

PAL=mpeg2source(palgout.d2v).assumefps("ntsc_film",sync_audio=true).killaudio()

NTSC=mpeg2source(ntscgout.d2v)

audiodub(PAL,NTSC)

Luke threw twice…maybe.

Author
Time
 (Edited)

Okay, it must be my TDecimate call screwing it up then, because the result jerks every few frames and doesn't sync... any suggestions for getting the frames at 24P?

Project Threepio (Star Wars OOT subtitles)

Author
Time

Yeah don't use tdecimate for this purpose. It's really for removing pulldown and you just need to do a pal->ntsc conversion.  Why 24p?  Even most 24p blu-rays are actually 23.976fps. If you want 24p it would be like this:

mpeg2source(palgout.d2v).assumefps("film",sync_audio=true)

Luke threw twice…maybe.

Author
Time

CatBus, I posted the scripts that you want in the first few posts.  Try those.

althor1138 said:Of course it would be better to encode the video and then mux in the audio later so as to avoid re-encoding but anyway here's how to do it in avisynth: 

PAL=mpeg2source(palgout.d2v).assumefps("ntsc_film",sync_audio=true).killaudio()

NTSC=mpeg2source(ntscgout.d2v)

audiodub(PAL,NTSC)

You don't need sync_audio=true or KillAudio() because MPEG2Source() processes only the video and because the script does not envisage using the PAL audio anyway.  Moreover, "NTSC" needs to point to an audio stream if you want to use AudioDub().

NTSC=WavSource()# or NicAC3Source() or whatever

But as you mentioned, the audio can be added on muxing.  All CatBus needs to do is apply the scripts given earlier in the thread.

Author
Time
 (Edited)

Whoah, that works, just AssumeFPS without TDecimate.  Slightly confused WHY this works to produce 23.976fps output, but willing to accept it on faith ;)

EDIT: Er, and thanks!

Project Threepio (Star Wars OOT subtitles)

Author
Time
 (Edited)

Could you guys give me a hand here?

I’ve installed Avisynth. Installed AVSP, which is the GUI for AVIsynth and I get tons of syntax problems with the scripts.
I’m trying to sync the video from the silver screen to GOUT.

I’ve tried creating an .avs file with the script in it and running it in windows media player like the AVIsynth wiki recommends but I also get tons of syntax errors.

Could someone tell me how do I apply the script to make my .h264 demuxed file to be saved with the script modifications?

EDIT:
It seems I’ve figured it out.
You need the following for it to work:
AviSynth
dgavcdec109
DGAVCDecode.dll plugin for avisynth
MeGui

The script in the end is this one:
loadplugin(“C:\Program Files (x86)\AviSynth\plugins\DGAVCDecode.dll”)
sw=AVCSource(“video.dga”).assumefps(“ntsc_film”).trim(1005,0)
blank=sw.blankclip()
swp1=sw.trim(0,42592) ++ blank.trim(0,11)#obi-wan reaches down to grab luke’s arm while he’s unconscious
swp2=sw.trim(42593,45140) ++ blank.trim(0,3)#luke and obi-wan sit on rocks by speeder while luke points at r2
swp3=sw.trim(45141,55729) ++ blank.trim(0,4)#cut from tarkin to r2/3po in front of sandcrawler----first frame of r2 is a duplicate(trimmed out) / end of reel?
swp4=sw.trim(55731,56064) ++ blank.trim(0,-1)#obi-wan / “that is what we are meant to believe”—first frame missing
swp5=sw.trim(56065,56904) ++ blank.trim(0,-1)#no luke, wait! / luke runs to speeder to go see owen and beru
swp6=sw.trim(56905,58453) ++ blank.trim(0,3)#tie fighters approach death star/cut to vader approaching leia’s cell
swp7=sw.trim(58454,60732) ++ blank.trim(0,-1)#standing at edge of cliff above mos eisley/cut to scene of mos eisley
swp8=sw.trim(60733,64836) ++ blank.trim(0,-1)#lined up at the bar/luke tugs on bartender’s sleeve
swp9=sw.trim(64837,70477) ++ blank.trim(0,1)#Han and Chewie stand up / Luke and obi-wan walk away from cantina
swp10=sw.trim(70478,78633) ++ blank.trim(0,-1)#Tarkin dismisses officer / Leia led down corridor by guards
swp11=sw.trim(78634,83416) ++ blank.trim(0,3)# Chewie with hands behind head / 3po looks at r2 and suggests a new strategy
swp12=sw.trim(83417,86622) ++ blank.trim(0,57)#That’s good. You’ve taken your first steps into a larger world. / doors open, officer enters in to deliver message to tarkin / End of reel
swp13=sw.trim(86623,113556) ++ blank.trim(0,5)#Dianoga releases Luke / Leia stares / End of reel
swp14=sw.trim(113557,143951) ++ blank.trim(0,4)#man in hangar waves droid into x-wing / overhead shot of x-wing in hangar / End of reel
swp15=sw.trim(143952,0) ++ blank.trim(0,5)
final=swp1 ++ swp2 ++ swp3 ++ swp4 ++ swp5 ++ swp6 ++ swp7 ++ swp8 ++ swp9 ++ swp10 ++ swp11 ++ swp12 ++ swp13 ++ swp14 ++ swp15
return(final)

Where video.dga is my video.h264 that where index by dgavcdec.

Open script in megui and adjust x264 settings:
I’m using CRF 16, very slow preset and default target device without min or max bitrate.

It will finish the encoding in 6 hours with my i5 4670k

EDIT:

I ended up without the first seconds of the movie.
What part of the script is removing it?
I’ve read somewhere that this was the case to remove the negative1 credits.

Which part is it?

EDIT:
Founded it

It’s this part: .trim(1005,0) <— this has to be removed.

Author
Time

Theory question here.

Is it possible to splice in the x264/h264 video of TN1’s SSE with the blank (or maybe registered) frames necessary to sync with GOUT-synced audio tracks…without re-encoding any of the non-blank frames?

I recall using Quicktime 7 pro to cut and paste together video clips that were the same resolution, codec, bitrate, etc…losslessly and selecting “pass through” when exporting to a new video file. I wonder if there’s a slicker way to do this losslessly because that was like 8 years ago (Pretty sure Quicktime 7 is free from Apple now, but I used Windows XP SP3 up until 2009 and have no idea if that’s true for other operating systems)

I imagine you’d have to demux the video track from the mkv and put it in an mp4 container before Quicktime could even open it, but I haven’t the system resources, hard drive space, etc to give it a shot myself. Figured I’d run it by the forum in case anyone here was working on a mac and hadn’t thought about it.

  1. encode blank frames to same settings as TN1
  2. demux SSE video to .x264
  3. remux SSE video into .mp4 container
  4. split SSE video losslessly into chunks that bookend where missing frames need to be inserted (no idea which program could do this on a PC or Mac, besides Quicktime 7 itself)
  5. merge those SSE clips with blank frames from step 1
  6. export with “pass through” in quicktime
  7. remux to mkv & test sync with the GOUT-synced until
  8. lossless re-hash of TN1 SSE

Thoughts?

Author
Time

a_o said:
Is it possible to splice in the x264/h264 video of TN1’s SSE with the blank (or maybe registered) frames necessary to sync with GOUT-synced audio tracks…without re-encoding any of the non-blank frames?

No.

Author
Time
 (Edited)

After having success with ANH from Negative One team and syncing it now I’m trying to sync ESB.
The problem is ConvertToYV12 which isn’t working properly.

While checking the video preview in MeGui the video is basically black and white.

I’m with Avisynth 2.6 which should support this.

Any ideas?

EDIT>
After a few hours reading and testing this solved my problem:

Loadplugin(“C:\Program Files (x86)\Haali\MatroskaSplitter\avss.dll”)
Loadplugin(“C:\Program Files (x86)\AviSynth\plugins\ffms2.dll”)
ESB=ffvideosource(“02-ESB_Grindhouse_1080p_BD25.mkv”)
.ConvertToYV12()#.ShowFrameNumber(scroll=true)

ESB2=
Trim(ESB, 0,3096).Loop(12,0,0) ++
ESB.Loop(6,30727,30727).Trim(3096, 35385) ++
ESB.Loop(3,35381,35381).Trim(35381, 59546) ++
ESB.Loop(7,59544,59544).Trim(59544, 59758) ++
ESB.Loop(5,59753,59753).Trim(59753, 75034) ++
Trim(ESB,75030,75978) ++
Trim(ESB,75978,90851) ++
ESB.Loop(7,90852,90852)
.Trim(90852,104633)
.DuplicateFrame(13781) ++
Trim(ESB,104627,121679) ++
ESB.Loop(24,121679,121679).Trim(121679, 150156) ++
ESB.Loop(11,150134,150134).Trim(150134, 0)

return ESB2

Author
Time

If you have FFMS2.avsi in your plugins directory, you shouldn’t need to load the plugin.

Author
Time

It was weird. I got all my Avisynth plugins from an outdated zip file that someone made.

So, I’ve updated the FFMS2 to the latest 2.22 version and I got a green upside down preview.
It was solved using the function to call just the video from the container.

I was hopping, initially, to use the AVC call command that I’ve used with ANH Silver Screen, but the AVC indexer software for the h.264 stream crashed all the time. I guess it’s not familiar with 4:2:2.