Red5

mverta said:

For what it's worth, I had information from a friend at ILM prior to the GOUT release that the original title and crawl were, indeed, recreated for the 2006 release, albeit carefully and conscientiously. 

 

I hadn't given it much thought until recently, when I noticed an extreme amount of jitter in the title card, which I have from two absolutely confirm-able 1977 sources, which are identical to each other.  What's odd is that the GOUT title card element is much straighter and smoother in its motion.  Stabilizing my 1977 elements doesn't yield the GOUT result; they have internally different characteristics.  If I didn't know better, I'd say either 1) GOUT was indeed a recreation or 2) There was a second, more stable version of the title card created after the initial release in 1977, but before the 1981 re-release and re-do, which GOUT is based upon.

 

You can view the differences for yourself in the Legacy forum, but there is more to the story here, for sure...

 

_Mike

I looked at your GOUT comparison and the Legacy title card element seems to drift quite noticeably more to the left than the GOUT and it seems kinda silly to mess up a simple thing like that if it was 'carefully' recreated in 2006. 

Also I compared the GOUT crawl with a sample from a Spanish SW 35mm print, it's only a one frame comparison but more or less from the same position where your clip ends, and it is a very close match, the Spanish title card element might be slightly slightly more to the left but not at all as much as the Legacy source and I think well within normal geometric distortions in the transfers/prints.

So my money is on the GOUT being the real deal and the Legacy crawl  "simply" is from another source.

Maybe the scene change scan buttons in Virtualdub can be used for this.

You can adjust the scene change thresholds in Options-Preferences for the level of change in the footage you want to detect and the scanning will stop there.

^@zombie84

What you're seeing is most likely the result of clipped white levels in the GOUT that is making the lightsabers core appear fatter. You can also see it in the ceiling light to the left, in the GOUT it contains much less white detail and nuances.

I don't have any of the 2004 DVDs available but here is (from top to bottom) the GOUT, the TB97 and the TB97 with white levels tweaked and clipped:

The electrical glowing effect in motion seems to be the same in the GOUT and TB97, if it appears different in the 2004 DVD maybe it's because the Lowry NR algorithms simply mistook some of it for noise and filtered it out.

No those frames are clean in the PAL (bottom) 13588

145590

Here's another error NTSC (top), PAL (bottom) 149900

Although here it is the PAL (bottom) version that's seriously messed up 148959

It's strange that the NTSC GOUT is missing frames compared to the PAL version as this is just a conversion from the same NTSC D1 mastertapes, but it probably happened when they joined the D1 segments and trimmed of any redundant frames and probably also messed up the IVTC.

There are i.e. interlacing errors in the NTSC (top) that are not present in the PAL (bottom).

A couple of frames missing at scene/reel ends really isn't a big deal though, except for perhaps loss of audio sync if anyone wants to use the DC/Faces PCM track with the GOUT video.

These two frames are missing from the NTSC ESB.

Darth Mallwalker said:

What's missing from GOUT



That frame is missing from GOUT



Here's those two fields after re-assembly

Faces frame from MeBeJedi prototype (IVTCed)


That frame shouldn't be missing from GOUT

But the PAL GOUT seems to have it.

Darth Mallwalker said:

Found more missing from GOUT

These frames from ROTJ:




It's the tail end of the establishing shot just before Mon Mothma's briefing.
Happens to be the end of a reel too.

Screen caps taken from Editdroid disc.
Now this is puzzling:
If Editdroid used DC laserdiscs, then 'DC master' should also contain those frames, right?
So what's GOUT's excuse for cutting them?
My childhood is raped!

Those two frames are intact in the PAL GOUT.

However the PAL ROTJ is missing this frame instead.

g-force said:

One quick suggestion, could someone replace the last line of the restore stars stage with:

MT_Merge(prest,source,last,chroma="copy first")

and let me know what happens?

thanks!

-G

yes, omitting the MT_Logic.. part from the restore stars stage does eliminate the artifacts.

as you probably already suspects the problem seems to come from MDegrain1 in this line:

pcorr   = source.MDegrain1(idx2,bw_vec1,fw_vec1,thSAD=800,thSCD1=150,thSCD2=75,plane=0)

pcorr has a lot of artifacts high motion frames and looks like this:

The lack of clipped white levels and different cropping seems to indicate that this LD certainly did not come from the 1993 DC/Faces/GOUT telecine.

Could this be the true 'shrink fixed' version davisdvd was talking about, but released in 1993 instead of 1992?

The 1989 laserdisc was quietly reissued with the newly corrected letterboxed transfer, completely doing away with the "incredible shrinking ratio" problem.

Jetrell Fo said:

LeeThorogood said:

 

austin_1138 said:

i downloaded it and when i tryed to unzip it it gave me an error and i couldnt extract it :(

I would suggest you try to download it again as the file may not have downloaded completely/correctly.

 

I had this same issue.  It said the zip was like 206 MB but the file it extracted was 0 MB....it may have been corrupted during upload too...I will try again as well.

my old Winzip 8.1 complained over invalid filename but maybe later versions can handle it.

anyway I used Winrar instead to uncompress it, no errors.

Bingowings said:

It could be that the act of photographing a cinema screen has made the images brighter/darker than they actually are too (depending on the conditions of each photograph).

Those two photographs were taken with the same shutter speed 1/30, ISO-1600, zoom and aperture, and from more or less the same angle and distance to the screen.

zombie84 said:

R2 canyon scene:

http://petergaultney.smugmug.com/Movies/historic/Star-Wars-at-The-Senator/K7PG2338/948667790_SRJST-M.jpg

Looks pretty dark!

Yes, but Luke looking at the sunset appear much brighter.

comparison:

http://originaltrilogy.com/forum/topic.cfm/What-changes-was-done-to-STAR-WARS-in-93/post/427767/#TopicPost427767

These two Senator theatre pictures seems to have been taken with the exact same exposure settings.

The R2D2 canyon shot seems to have more or less the same brightness level as the GOUT but the shot with Luke looking at the sunset is clearly much brighter and with higher contrast.

The question remain though if it was a 'deliberate creative decision' to make the sunset appear darker in the GOUT (1993 telecine) or simply the effect of the colourist trying to hide the defects from a noisy and faded print, which is what I tend to believe.

As far as I can see the framing in the rebel shot also seems to match the NTSC GOUT.

Also keep in mind, if you're comparing the ED2 with the PAL GOUT, the NTSC GOUT is slightly sharper and more detailed than the PAL version and as already mentioned differently cropped.

ED2 top, GOUT bottom

I think there might very well be more advanced video processing involved than that, but it's difficult to speculate with just these screenshots and without having seen it in motion.

The source really seems to be the NTSC GOUT DVD though.

zombie84 said:

Moth3r said:

From closer examination of the screenshots, I believe this is actually another GOUT-sourced Custom DVD, and not a laserdisc capture at all.

 Why is that?

Also, if they had taken the GOUT and re-worked it, why wouldn't they just state that, as that would, on the surface, seem like a more sensible solution to most people?

The screenshots are framed exactly as the NTSC GOUT DVD although upscaled to 16:9, and they show the exact same amount of picture information.

The known LDs (faces and DC) have slightly less picture information left and right and slightly more up and down compared to the GOUT.

Maybe the term "LD source" or "LaserDisc version" have been somewhat confused by LFL GOUT 'DVD' release as they used the old 1993 master that was in fact only intended for LD and VHS releases.

ED2 top, GOUT bottom

ED2 top, GOUT bottom

272-276 lines (pixels) seems to be a common size for a letterboxed NTSC source and would be defined in the telecine process. And for a LB PAL source it would be 326 lines (pixels). And these numbers should remain the same even if the source was transfered to another format, without any vertical resizing of course.  

Comparing with the ROTJ Faces LD it seems both the PAL and NTSC GOUT have been vertically resized and cropped. The LD is 276 lines high, the NTSC GOUT is 274 and shows less information both top and bottom more less than the 2 lines difference. The PAL GOUT is 326 lines high and shows the same amount of info at the top as the LD but is cropped 7-8 lines more at the bottom.

Only LFL knows why it was done this way, there's really no technical reason for it. Making the PAL GOUT from the NTSC source only require a simple vertical resize from 276 to 326 lines and adjusting the black bars to fill up the 576 lines of a PAL DVD picture.

The PAL and NTSC GOUT are definitely from the same masters (the 1993 NTSC telecine), but aside from the NTSC to PAL resize and slightly different cropping, it seems the PAL ROTJ was processed differently and had less (or none) vertical blur filter applied, there's clearly more vertical detail visible.

The NTSC ROTJ also has more  'ringing' artifacts like extra 'lines' around horizontal lines and contours, probably a result of to much vertical sharpening.

you can also look for Arnie.d's MU links in this thread.

I think there are 64 ANH samples, 29 ESB and 6 ROTJ.

scaling the gout to the 70mm samples the gout appear more cropped.

Well comparing the vertical details from the X0 project and another X0/X9 LD capture, in these screenshots the GOUT appear to have the same vertical resolution as the LDs.

GOUT

X0 project

GOUT

other X0 or X9 capture

But I agree there are some other shots/scenes/areas in the GOUT where the vertical resolution seems to be cut in half, but I would suspect it was caused by a crude AA filter and to some degree incomplete (lazy) IVTC work.