THX Italian 35mm - Feedback thread

Great idea! I’d like to compare/align the 35mm IB Tech (Italian) & the 16mm scope (English). [If we have any “official” names for these projects, can someone list them? I don’t know what to label anything.]

I’m using Avisynth to align everything (it’s all manual, so don’t think computers make things easier). Aside from coding and debugging the code, I’ll watch and try to sync the video streams, and point out missing frames and changed edits … roughly this way:

The 35mm shows a jump in the action (THX's 1st time in his medicine cabinet) in the top row. The next rows show the 16mm frames going out of sync and coming back into sync. All frames are numbered from the start of their respective files. So a little math will reveals how much is missing and where.

I’m working towards 2 film strips that will play with placeholder frames as the sync changes to maintain sync between them to the end. That will take a while, so I’ll put up pictures like the above just to get the info out as soon as possible. So, anyone else doing this, don’t wait for me. Just post your findings, too.

It’s slow and I’m normally short on time (not to mention all of us, but that’s another story). Anyone – please duplicate my effort your way because poita needs such information to meet his more crucial deadlines.

I’m using the 16mm scan as a base and aligning the 35mm to it. One problem I came across is the 16mm has duplicate frames here and there. I’ll check out the full-sized (I work reduced-sized for reduced-lagged) to see if the frame code increments on the dup’ed frame, too. And as you mentioned, the 35mm has groups of frames missing … just to make things more fun. 😃

BTW, what software are you using?

I was going to show the results of this Avisynth script to align the 2 scans. On double-checking before posting, I noticed the video of one film was shifted by 1 frame even though none of the settings had changed (and were printed onto the picture). Going back and running it again, they then were properly aligned, with the same settings . .

Clearly something in Avisynth (this system is running v.2.5.8) is wrong and cannot be relied upon to serve up the correct images. Sorry this approach turned out to be useless.

(Normally I’d put a cool, animated smiley here, but my smiley resource site "http://www.freesmileys.org/" completely disappeared a several days ago …

… and a lesson for us all -[as spoken by Tony "Felix Unger" Randall].) 😉

Well … after looking around for something to replace Avisynth and finding programs only useful for editing video, I’ll just move along with Avisynth with the caveat to look where I found missing areas and verify them independently.

In the first 1,000 frames of the 35mm, I found these:
[UPDATED for fixes & presentation:]

sync 35mm to 16mm - as 2 parallel streams matching where they can
1st 1000 frames

35mm frames . . description
00000-02265 . . start: 35mm-Italian (titles; missing counter & medicine-spycam)
. 00000-02973 . spacer for missing 16mm titles & counter & medicine-spycam
02266-02330 . . begin match @ medicine-spycam
. 00000-00011 . spacer for spliced-out medicine-spycam
02331-02377 . . begin match @ medicine-spycam
. 00000-00000 . spacer for spliced-out medicine-spycam (1 frame)
02378-02647 . . begin match @ medicine-spycam
. 00000-00004 . spacer for spliced-out print-out
02648-05027 . . begin match @ print-out
. 00000-00006 . spacer for spliced-out control-room
05028-05345 . . begin match @ control-room
. 00000-00000 . spacer for spliced-out manipulator (1 frame)
05346-07297 . . begin match @ manipulator
. 00000-00044 . spacer for missing 16mm radiation & sparks
07298-07340 . . begin match @ sparks
. 00000-00000 . spacer for spliced-out sparks (1 frame)
07341-07670 . . begin match @ sparks (NOTE: frame wobble)
. 00000-00003 . spacer for spliced-out THX-adjusting
07671-09218 . . begin match @ THX-adjusting

16mm frames . . description
. 00000-02265 . spacer for 35mm-Italian titles
00000-04305 . . start: 16mm-English (titles)
04307-04341 . . begin match after skipped duplicate frame #04306
04343-09961 . . begin match after skipped duplicate frame #04342

(I just realized that I should’ve treated all Italian-language video-changes like the opening titles. Will redo the “radiation” section and edit here.) Done.

Spaced Ranger said:

I was going to show the results of this Avisynth script to align the 2 scans. On double-checking before posting, I noticed the video of one film was shifted by 1 frame even though none of the settings had changed (and were printed onto the picture). Going back and running it again, they then were properly aligned, with the same settings . .
Clearly something in Avisynth (this system is running v.5.8) is wrong and cannot be relied upon to serve up the correct images.

I rather doubt it. It sounds more like a decoding problem. What are you using to decode these videos and what format are they in?

By 5.8, I assume you mean 2.5.8. Try upgrading to the latest version. I prefer to stick to the 32-bit non-MT version.

I know, and I was stunned at such a problem surfacing. (And, yes, it should’ve been v.2.5.8 . Been having typos lately. 😃 I’ll correct it above.)

Thanks for the things to consider. The video files were originally extensioned .mov (which wouldn’t read), but a quick web-search suggested changing the extension to .mp4 (which then did read). I had a few versions installed/removed but don’t recall installing anything special for file reading. Also had v.2.6.0 installed once to try something from OT (with a few recommended plug-ins), but it wouldn’t work the script as it should’ve. (Sometimes I hate Avisynth.)

I believe it!

<rant on>
In fact, while rerunning my script after working in/out the Italian-changed text in the movie, I received an inexplicable “I don’t understand” error about one variable, which was called once. Impossible! After trying different things (without any success), I change one character in the name (in both places), and, inexplicably, it runs again. Worse, I change that one character back to it’s original character (in both places), and now it works when it didn’t before! Grrrrr!!

And then there’s always those testimonies of those who try to run scripts (which most people run just fine) but can’t get it to work.
</rant off>

There. I feel better now. I guess the next thing to check into would be memory integrity (shudder).

Spaced Ranger said:
I believe it!

I don’t.

AviSynth can decode MOVs with QTReader or QTSource. Try those. It’s been a while since I’ve used either. (I avoid MOVs like the plague.)

Also had v.2.6.0 installed once to try something from OT (with a few recommended plug-ins), but it wouldn’t work the script as it should’ve.

That’s rather vague, so I’m not sure what to suggest. While bugs are always possible, AviSynth is heavily used by some very knowledgeable people, so bugs tend to come to light. The frame-shifting problem that you’ve described sounds like it could be a decoding issue, and I’d sooner attribute it to that than to something fundamentally wrong with AviSynth’s ability to serve frames accurately. If you’re serious about investigating the problem, consider posting a thread over at Doom9 about it.

I guess the next thing to check into would be memory integrity (shudder).

Yes. That might be a good idea.

I’ve update the 1st 1000 frames post (above). Suggestions (especially from poita, as he’ll be referencing this) for better clarity and more useful formatting are welcome. Otherwise I’ll continue this format.

ADDED:
@ Mavimao

Don’t know if you’ve logged all your match-edits, but does it look like I’m catching what you’ve already found so far? (And if you have a log of matches, would you post them here? I’m using both frame-counts and SMPTE, so in either format is workable.)

It seems less splices as we move into the reel:

sync 35mm to 16mm - as 2 parallel streams matching where they can
2nd 1000 frames

35mm frames . . description
09219-15933 . . begin match @ mall-hall
15934-16074 . . begin 35mm-Italian LUH-report
. 00000-00145 . spacer for 16mm-English LUH-report & spliced-out LUH-cabinet (2 frames)
16075-19191 . . begin match @ LUH-cabinet

16mm frames . . description
09962-16676 . . begin match @ mall-hall
. 00000-00140 . spacer for 35mm-Italian LUH-report
16677-16820 . . start 16mm-English LUH-report
16821-19939 . . begin match @ LUH-cabinet (-2 frames)

Clutchins, please stop posting that.

LOL

Okay, okay … It seems less fewer splices as we move into the reel.

Well, major slow-down at the “malfunctioning officer” shot. Both 35mm & 16mm have missing frames, much of the area of interest is the walking-away-from-the-camera character (little movement to compare), and … someone putting his 16mm counters in picture (industry standard?), right over the walking legs.

Help me, Obi-Wan AntcuFaalb. You’re my only hope … you and your THX laserdisc capture, that is (is finished?). 😃

If you find time and if this helps, here’s a short Avisynth script to put one, or the other, or both numberings on the frames. If iMovie can import/handle .avs files, you could watch it from your timeline. Otherwise, you might be able to view it in a videoplayer:

file name (completely arbitrary): THX_numbered.avs

##=================
## video w/ frame numbers
##=================

## Video path & name
## ----------

movie = "THX1138_35_stabilized_HD\THX1138it.mov.mp4"

## Get video & enforce format
## ----------

film = DirectShowSource( movie )
\ .ConvertToRGB().ConvertFPS( 24 )

## Add counters
## counters slow the display
## remove "#" from only the one to show
## ----------

#film = film.ShowFrameNumber( x=100, y=50 ) # frame number
#film = film.ShowSMPTE( x=100, y=75 ) # SMPTE
film = film.ShowFrameNumber( x=100, y=50 ).ShowSMPTE( x=100, y=75 ) # both

## Display
## ----------

film

Hi, Can someone pass to me the italian dub - only the audio track - synched to the laserdisc?

I just found out now about this thread, sorry for the lateness. I’ll be checking your progress and I’ve had so little time that I didn’t actually get to have a proper look at the film yet.

robbottin, I don’t think anybody has synched the Italian dub to the laserdisc yet (I don’t even think anybody was planning to, to be honest), why do you need it?