logo Sign In

Thought on de-SE'ing the DVD — Page 3

Author
Time
Post link
Big news on the green saber issue.

This evening I installed the demo for Combustion, based on MBJ's rave reviews. Though I'm very comfortable with Photoshop, I was lost with the Combustion interface. But, with most software packages, I'm like a duck in water if left in the pond long enough.

So, I made a sample rip of the DVD ... specifically the Falcon sequence with the infamous green saber.

I was planning to repaint the thing by hand or whatever it took. Then, while randomly hitting buttons with very little thought, trying to figure out the interface, I found a "Discrete color correction" utility in this program, with a curious set of buttons under the heading "rewire." I started playing with these options and found if you changed these two settings

G <- Y
B <- G

The saber is fixed, and the rest of the shot isn't screwed up. That's right!!! Nothing appears wrong (at least, to my untrained eyes) when I "rewire" the composite data in this fashion. Flesh tones look fine. Color balance looks right. NOW tell me that somebody didn't screw up at Lucasfilm. This is the kind of "crossed wires" mistake that the miswired music in the rear channels reeks of. I am 100% convinced now that this is no "deliberate creative decision" as they're purporting. I did absolutely nothing else to the shot. I just mimicked the composite rewiring in said fashion, and the bad "green saber" shots are now a gorgeous blue, without sacrificing the rest of the shot. I'm sure this will screw up other shots in the film that aren't messed up, but this is very, very, very important information. If we can correct the bad shots, then we can use a utility like the software I mentioned a few posts above to splice in the corrected shots.

This photo is a direct output from Combustion. I did nothing from the source video other than rewire these two colors. The output was a TIFF, which I then converted to a JPEG in photoshop for easier posting to the web. That is all. If the shot is too dark or too saturated or whatever, well, that's what some are saying about the disc as a whole, but this matches the surrounding footage, does it not?

Link removed and put in a post below.
I am fluent in over six million forms of procrastination.
Author
Time
Post link
The only way to be sure would be to post the orinal frame as well for a compare.
Basically you are remapping the colours, so anything that was green in shot will now be blue. If this was an optical printing error originally (i.e. they optically screwed up and ended up with green rather than blue) then this would actually correct the anomolies in the whole frame. If not (i.e. they just coloured in the sabre green by mistake) then you are adjusting the colour balance of the shot away from the original.
There is hardly any bright green elsewhere in the scene, so you could probably get away with it anyway.
Author
Time
Post link
Please excuse any mis-typings, I'm doing it in the air, and its a bit turbulent at the mo'
Author
Time
Post link
The guy's giving advice from the seat of a plane. How cool is that?

Anyway, Laserman's absolutely right. You may not be able to tell if swapping the green and blue affects anything other than the saber. Though I have to say that it still looks a bit light. But I'm sure there's a way you can adjust the saturation of just the saber.

My Projects:
[Holiday Special Hybrid DVD v2]
[X0 Project]
[Backstroke of the West DVD]
[ROTS Theatrical DVD]

Author
Time
Post link
I've played with the settings a bit more, and the B <- G may even be unnecessary. The B <- B brings out more detail in things like R2, but less in the Saber. I do think this is a screwed up color correction setting and not a miscolored saber any more. How this got past QC still escapes me.

Okay, here are the screencaps. I said I was too tired, but here they are anyway. Now I'll be able to sleep.

Unretouched shot:
http://216.12.134.120/unretouched.jpg

Posted earlier: G <- Y and B <- G
http://216.12.134.120/y2g-and-g2b.jpg

Alternate version: just G <- Y
http://216.12.134.120/y2g.jpg
I am fluent in over six million forms of procrastination.
Author
Time
Post link
Well, obviously in the second shot, the blue has turned green on R2-D2 and on the 3 panel lights at bottom of shot, Luke's pants are no longer sandy/olive, the chess pieces change colour and many other colour problems appear. So the original re-wire does fix the sabre, but screws up the colour in the shot altogether.

The last shot is more subtle, but still mistimes the entire frame.

Use the selection tool in combustion to just select the sabre itself - you will need a travelling selection for a frame sequence and just recolour the blade. It looks as if the rest of the frame is correct in the original shot.
Author
Time
Post link
something occurs to me here. BTW, adigitalman, that is a sweet post. I'm thinking of toying with combustion myself and the more I hear, the more I think this is going to be the program of choice.

Lowry digitally scanned all the film right, and this is after LS monkeys color corrected it. how's about they actually did something partly right during this 'restoration'.

what if they went back to the original source stuff that had previously been restored, color corrected it and this includes the optical composite elements. only they then scrwed up and someone wasn't watching when they were doing the lightsaber rotoscope mattes.

it might explain why just the saber is screwed for color in some shots and the rest of the scene is fine. if they had color corrected a finished fully composited frame, then it would be logical to assume that other colors would be off as well. the fact that only the saber color is wrong in quite a few shots suggests that they did something else and this is why we might have to manually re-rotoscope the sabers or do a travelling selection to just alter the saber color.
When a woman says yes, she means no - when she says maybe, she means no.

http://www.auky37.dsl.pipex.com/falconlogo_web.jpg
Author
Time
Post link
Laserman, thanks for a more critical eye. I see some of the things you're talking about, and frankly just wouldn't have if I hadn't asked some of you real videographers. I am, after all, not experienced at all with video editing. Can you PM (or post here if you think it's not a waste of everybody's time) with instrux on how to do things like selecting just the saber? I couldn't figure this out. I was hoping for somethign as user-friendly as photoshops magic wand tool or lasso tool.
I am fluent in over six million forms of procrastination.
Author
Time
Post link
There are so many ways you could do it. You could roto a matte for the sabre, you could use the keyer to pull a matte, you could use a travelling selection, you could hand paint each frame etc. etc.
If you want to get an intro to combustion there is a simple and quick tutorial here
http://www.digitalvideoediting.com/2004/03_mar/tutorials/ceyeaeguy1.htm
and buckets of others on the web
Author
Time
Post link
I'm starting to think this thread belongs in a different forum, but we're pretty deep into it now, so I apologize and ask you to bear with me. I'm back on ROTJ for the past couple of nights, working on Lapti Nek. I wanted to do something totally bizarre based on our limited source material for the audio, and the problems of mismatched video quality. So I've done something unique to my "reduced special edition" ...

Okay, so it's no secret how badly I hate "Jedi Rocks." Having played "Lapti Nek" for five years in marching band at Friday night football games as a kid really endeared me to the song. Plus, "Jedi Rocks" is quite simply a horrible, horrible song. Unfortunately, the only way to restore "Lapti Nek" is to re-insert a good capture from LD, and there is still the concern of a mismatch in video qualtity, audio levels, etc. To boot, the film version of "Lapti Nek" isn't available anywhere to my knowledge. So I'm creating something totally unique, that's a bridge of sorts between the OT and the SE. There will be enough good OT restorations to sate our taste for the OT, so I'm taking an approach that's completely unique.

I took the CD version of Lapti Nek from my Anthology CDs and reprocessed thru a 5.1 encoder. I'm not sure how well this is going to translate in the end, but I'm trying. I then took the entire "Jedi Rocks" DVD video and cut up into the indivudal shots. I found a reasonable start point for the song, at the first horn flare, since that's how "Jedi Rocks" begins. I then began to reorder and match as closely as possible, different shots from the SE, up until the point where Oola is dropped into the Rancor pit. A handful of shots are re-used, but I tried to use different frame sequences where possible. The video is pretty much set to the music, though I may tinker with one more shot. I'm working now on Oolas struggle with Jabba, trying to get in some "dialogue" into those shots. I was able to sample audio elsewhere of Jabba talking that works for his vox, but I need to find some audio that will work for Oola. Obviously, the DVD audio is out, as it has music underneath that clashes. I return to the original audio with her scream as she goes down the pit (there is still a bit of source music there, but the crossfade seems to be working nicely), but I need a few grunts and "nos" leading up to that to make the whole sequence work a bit better. Any ideas?

I am fluent in over six million forms of procrastination.
Author
Time
Post link
//grunt

Better now?

<span class=“Italics”>MeBeJedi: Sadly, I believe the prequels are beyond repair.
<span class=“Bold”>JediRandy: They’re certainly beyond any repair you’re capable of making.</span></span>

<span class=“Italics”>MeBeJedi: You aren’t one of us.
<span class=“Bold”>Go-Mer-Tonic: I can’t say I find that very disappointing.</span></span>

<span class=“Italics”>JediRandy: I won’t suck as much as a fan edit.</span>

Author
Time
Post link
Welcome back, Master MBJ. It hasn't been the same without ya.
I am fluent in over six million forms of procrastination.
Author
Time
Post link
Recognize the pics in my .sig?

<span class=“Italics”>MeBeJedi: Sadly, I believe the prequels are beyond repair.
<span class=“Bold”>JediRandy: They’re certainly beyond any repair you’re capable of making.</span></span>

<span class=“Italics”>MeBeJedi: You aren’t one of us.
<span class=“Bold”>Go-Mer-Tonic: I can’t say I find that very disappointing.</span></span>

<span class=“Italics”>JediRandy: I won’t suck as much as a fan edit.</span>

Author
Time
Post link
Indeed I do. You should animate them with the fade back-and-forth that I saw on the one file.
I am fluent in over six million forms of procrastination.
Author
Time
Post link
Well, as soon as Rikter can fix his FTP site (hint, hint), I'll post the MPEG for all to see.

<span class=“Italics”>MeBeJedi: Sadly, I believe the prequels are beyond repair.
<span class=“Bold”>JediRandy: They’re certainly beyond any repair you’re capable of making.</span></span>

<span class=“Italics”>MeBeJedi: You aren’t one of us.
<span class=“Bold”>Go-Mer-Tonic: I can’t say I find that very disappointing.</span></span>

<span class=“Italics”>JediRandy: I won’t suck as much as a fan edit.</span>

Author
Time
Post link
About color correction...

Every video transfer get's its own color timing, every newly printed interpositive get's its own color timing,
so what do you consider the original color timing then?

Every time you run the film through telecine machine,
you have to color correct the video data because film images are not really "compatible" with video,
so you need serious contrast corrections and color corrections. And there has to be someone
there to supervise the transfer, I supose Lucas was there every time (I don't think he would let anyone else
fine tune his films) and he made corrections.

The thing is that the raw interpositives have an orange low con image, and to view them you have
to correct them. Telecine machines have a default way of correcting this, and sort of emulating
print film contrast. But still there are numerous ways of rendering this raw footage.
There are 4 "flavours" for cinema print film today (2 Kodak, 1 Fuji, 1 AGFA) , each one has it's own
contrast, color etc. Therefore, not even direct one-light prints from a negative are objective.
And there have been a lot of print films in history, each with different look to it.
And as for video transfers, things are even more relative and subjective.

There IS NO "original" color balance and exposure setting.
The negative has over 10 stops of exposure, if you show all 10 you end up with a very very low contrast washed out image,
you can choose how many stops to show, and which stops to show.
The negative has an orange mask, and all the colors are way off to cyan-ish, a cyan-blueish nightmare to be exact.
So chose your color balance because the original color balance is unwatchable.
There IS NO standard way of rendering this raw footage. Each print film has it's own
way, and each telecine transfer is a matter of choice and taste.

I guess if you consider the original color balance and contrast, that which was in the
prints in 1977, then you are in trouble because where will you find a non-faded SW
print today, or a SW print at all. The only surviving evidence of original color timing today
are a few Technicolor prints made in 1977, Lucas has one, Gary Kurtz had one etc.
These never fade and are a good color timing reference.

But video transfers are very subjective by nature, and never show what is
on film really. video transfers are "subjective interpretations of film images" so to speak.

All you can do really is correct the images on your own taste, because there is no
way that you can see the original timing today.
Author
Time
Post link
Cubebox, that's very informative. Some of what you wrote is about what I thought I understood from reading about telecine stuff, but I didn't know. And of course you've added some more detail that I've never seen elsewhere.

(BTW, you want "gets", not "get's"; and "its" everywhere, not "it's". )
Author
Time
Post link
Thanks for the grammar corrections, English is not my local language...

Anyway, to clear the things a bit more. This is a very popular analogy:
Take your SLR film camera, shoot the same subject under sam light several times
using different exposure setting on the camera, each time increasing the exposure by one stop
(open the iris by one stop). Give it to the lab, they will process the negative and
make prints out of each of these frames. All the prints are going to be identical in
sense of exposure. The minilab printer meters avarages every exposure on the paper.

The negative has a very wide range of tones, and it is up to filmmaker to decide
what smaller range is he going to extract out of the negative for his prints.

For example, you shoot a scene of a man in a room filled with daylight comming through the windows.
The difference between the shadowy interior and a sunshine lit exterior is HUGE.
You are exposing for the interior details.
The negative will record the interior details as well the exterior seen through the window.
So if you print that scene to print film for projection at normal printer setting, you will
get the image as you exposed it: white windows with almost no detail, and well exposed interior
showing normal colors.
But if you set the printer to a different setting, you can have a black interior with no detail,
and a normaly exposed exterior showing the scene outside the window with normal colors.

So basicly , you could say that you can "squese" two normal (viewing contrast) images into
one image on the negative. The range is that big.
In other words, the original negative holds a LOT of brighter tones than the brightest tone
on the cinema print (pure white).

That is the main reason for all the lightsabre troubles in the new DVD's. Some people think
that they recolored the cores so that they are pink rather than white now.
They didn't, they just used a different "printer setting" for the original image.
The pure white you see on the screen is not the brightest color in the negative.
So a film transfer is just a matter of chosing your white point and your black point in
this greater range.

The new DVD's are "printed down" a bit, the white point has been moved a lot higher,and
because of that you can see a lot more detail in the white corridors, bright sabres and other
bright tones that you did on previous transfers. This is basicly a good thing because
you can always go from higher dynamic range to lower (more punchy blown out highlights) by
making your own adjustments, but you can't go the other way around.
You can't retrieve a blown out highlight from a video file.

I think people expect the original "elements" (as they put it in articles) to be like some sort of
slides that you can look at on a lightable and see the "right" color balance and contrast.
But if you broke in the Lucasfilm vault, you would find unusable orange, ultra low contrast
images that require color timing and show nothing about the original color balance.

So what I was really trying to say is that people who are trying to restore the trilogy to its original
look need to be more specific about what do they consider to be "original" color and contrast.


Here is a perfect example:
If you took the original camera negatives of the scenes in Jabbas palace or sandcrawler interior
and scan them on a film scanner, you would not get what you see in cinema or video transfers.
Instead you would see a brightly lit scene where there were hardly any deep shadows.
All those sets were lit very brightly (it was 100ISO film for ANH and ESB) with strong lights,
and exposed as that. The images were then darkened in post production in the stage of color timing.
It was Lucas's creative decision. The Jabba's palace looks dark in the preprint copies, and in the prints.
But it is bright in the originals. Film is exposed for shadows usually, so what you see as shadows
in the print were actually normal mid tones on the set and in the originals.

So the question of what is original is quite a phylosophical issue.






Author
Time
Post link
Interesting info. I guess with regards to this thread, the color timing of output video would need to match that of the DVD itself, given that the idea is to take footage from other sources and splice it in place of the SE content so that the video and audio match as closely as possible.

I'm interested in more on the color timing and whether the green saber in ANH merits being called a mistake as I and others believe. Conceptually it is a mistake, as we know the saber is supposed to be blue story-wise. But in the vein of your points about color timing and shifts during digitization, was this saber re-rotoscoped, or did the shift happen in the simple transfer process of film-to-digital? And if so, why would this be allowed by LFL? To me it looks like a glaring error that inexplicably escaped quality control.
I am fluent in over six million forms of procrastination.
Author
Time
Post link
I have closely examined the screenshots from both versions (SE and 2004 DVD).
All the little textural variations of the glow "field" are identical, it is the same glow as it was in SE.
The original glow was allso greenish, but with a lot less saturation and more leaning to cyan.
I believe that they corrected the sabre locally by boosting its saturation and correcting it to a more pure green.
But I am sure that it was not a newly repainted glow, it's the same old glow, but with a new
color.

Anyway, it has nothing to do with scanning or whatever, the cyan-green glow was there from the begining,
it was just not as pure as now.
Author
Time
Post link
Of course colour is a subjective process, but you can tell when it is just plain wrong also.

If people's faces are bright orange, and white walls are leaning towards blue then you know the process has gone off the rails. As for setting the greyscales/gamma then that is totally subjective - In theory you want all levels of grey clearly visible, but a creative decision might be to darken up the shadow areas for 'mood' and so lose detail in the shadows that was there on the original film.
Also, a CRT cannot display the same colour space as film, so they are radically different.
Author
Time
Post link
came across something when viewing the 2004 SE DVD extras last night that we may find useful.

hint. watch the 1997 re-release trailer. it's been sharpened up but the color values haven't been messed with.

ie, we can use it as a comparison shot to establish color values on the SE DVD to get it back to how it should be. the 1997 trailer has no blue caste to it at all.
When a woman says yes, she means no - when she says maybe, she means no.

http://www.auky37.dsl.pipex.com/falconlogo_web.jpg
Author
Time
Post link
Originally posted by: Laserman
Of course colour is a subjective process, but you can tell when it is just plain wrong also.

If people's faces are bright orange, and white walls are leaning towards blue then you know the process has gone off the rails. As for setting the greyscales/gamma then that is totally subjective - In theory you want all levels of grey clearly visible, but a creative decision might be to darken up the shadow areas for 'mood' and so lose detail in the shadows that was there on the original film.
Also, a CRT cannot display the same colour space as film, so they are radically different.



True to life colors are not always a good measure either.
The balance is not always intended to be towards white. (remember matrix)
Take hoth blizzard scenes for example, they are blueish, yet it was shot in
day where there is enough light.
They couldn't possibly shoot it with 100ISO in the early evening in the middle of a storm.
There just isn't enough light in artic storm conditions to shoot it in the evening. It had to be
white day light to get enough exposure.


Author
Time
Post link
Originally posted by: tellan
came across something when viewing the 2004 SE DVD extras last night that we may find useful.

hint. watch the 1997 re-release trailer. it's been sharpened up but the color values haven't been messed with.

ie, we can use it as a comparison shot to establish color values on the SE DVD to get it back to how it should be. the 1997 trailer has no blue caste to it at all.



This is what my point was all about.
Rather than say "haven't been messed with" you should say "have been messed the same way as before"
Film images are ALWAYS messed with one way or another.
Author
Time
Post link
Just a little something I ran across years ago.

SIDEBAR: The Re-Emergence of Technicolor™

Dye transfer printers are elated at recent indications that Technicolor—the cinematic version of dye transfer—is returning to the big screen. The re-release of Giant in 1996, the first American-printed Technicolor feature film in 21 years, has heightened interest in its revival within the film industry.

Technicolor is also called IB printing (for “imbibition”, after the photographic term “dye imbibition”). Technicolor, Inc., ended IB printing in the U.S. in 1974. Technicolor London closed its operation in 1977, but not until they’d made five IB prints of Star Wars for George Lucas.

The restoration of the Star Wars trilogy brought IB printing back to the forefront. According to Leon Briggs, who worked with Lucasfilm on the restoration for over two years, the original negatives had faded only 5 - 15%, well within normal range. But he explained that George Lucas wanted the original color in the restored version. Lucasfilms technicians were able to accomplish this goal for Star Wars, but only because they had an IB print to use for color reference.

There is no good reason for the DVD colors to look the way they do, especially in light of this and the cost of each print. Terrible, terrible, terrible.

<span class=“Italics”>MeBeJedi: Sadly, I believe the prequels are beyond repair.
<span class=“Bold”>JediRandy: They’re certainly beyond any repair you’re capable of making.</span></span>

<span class=“Italics”>MeBeJedi: You aren’t one of us.
<span class=“Bold”>Go-Mer-Tonic: I can’t say I find that very disappointing.</span></span>

<span class=“Italics”>JediRandy: I won’t suck as much as a fan edit.</span>