logo Sign In

Remastering the 1981 Episode IV Title/Crawl/Flyover (Released) — Page 3

Author
Time

Goodness gracious…I don’t envy you.

A question for you yotsuya, Is the only color in either the '77, '81 or 97 starfield, Black and White, or is there variations of Blue in there somewhere?

Author
Time

I have 5 sources. The 1982 PAL release (the crawl/flyover was horizontally compressed rather than cropped), the 1985 US Special Widescreen Edition, the Japanese Special Collection, the 1995 Definitive Edition/Faces Edition (the same source as the rest of the GOUT), and the 35 mm scan that was included with the Silver Screen Edition. Not one of them is good enough to even match the GOUT. The 1982 version has a lot of horizontal smearing (The stars are ovals). The words in the LD versions have artifacts that distort the edges, not to mention some ghosting. Some are too bright and you can see the full disc of the moons and the edge of the atmosphere in the matte, which, like the effects mattes, should be dark and invisible. Then you have the film scan which has super crushed blacks. The few stars that are visible are tiny specks.

As the 1981 crawl/flyover is the one I grew up watching at home, it is the one I want most to go with either the restoration from film, a despecialized version, or an enhanced GOUT. So I am willing to spend the time to make something that is close enough to the original from matching original elements to get it. I’m doing it in 1080p so it will match the 77 35 mm crawl/flyover in quality. It is a challenging but fun exercise.

Author
Time

dlvh said:

Goodness gracious…I don’t envy you.

A question for you yotsuya, Is the only color in either the '77, '81 or 97 starfield, Black and White, or is there variations of Blue in there somewhere?

The only thing I have to go by is the enhancement that Poita did. He used astronomy software to pull out a lot more stars than I could detect out of the 35 mm scan of the '81 starfield and a lot of them have some color. I went under the assumption that the color he came up with was oversaturated because none of the other shots of the same starsfield (it was created for TESB and various pieces of it show up in like 50+ different shots in TESB) don’t show any color to speak of. When I up the saturation, the small stars tend to go blue and the large ones tend to go red. But using the original saturation they all tend to be neutral in color - white or gray.

Author
Time

yotsuya said:

dlvh said:

Goodness gracious…I don’t envy you.

A question for you yotsuya, Is the only color in either the '77, '81 or 97 starfield, Black and White, or is there variations of Blue in there somewhere?

The only thing I have to go by is the enhancement that Poita did. He used astronomy software to pull out a lot more stars than I could detect out of the 35 mm scan of the '81 starfield and a lot of them have some color. I went under the assumption that the color he came up with was oversaturated because none of the other shots of the same starsfield (it was created for TESB and various pieces of it show up in like 50+ different shots in TESB) don’t show any color to speak of. When I up the saturation, the small stars tend to go blue and the large ones tend to go red. But using the original saturation they all tend to be neutral in color - white or gray.

Thanks yotsuya.

How goes the progress?

Author
Time

I completed a test of overlaying the engine glow. On one hand it worked very well and on another it was iffy. I think I need to tweak it some, but to do that I need to finish the flyover. The good thing about procrastinating is that sometimes you think of a better way. I’m going to be using the 1977 flyover after I make sure all the details match the 81 flyover. I suspect there was some difference in matting some of the effects. As the blockader runner disappears behind the Star Destroyer, the engine glow of the BR shows a matte line in the 77 version and doesn’t in the 81 version. And I thought I saw a laser bolt that was different. So as soon as I correct for everything like that, I’m going to start making my layers. I found a way to cut out the stars, planet, and moons so I can make a matte. It will require some tweaking to remove the lasers, engine glow, and any stray stars along the edge of the ships. But once I have the matte, then it is fairly easy to get the ships isolated. The way I found of making the matte also gives me fairly decent laser and engine glow elements. It is a lot less work than what I first had in mind.

So I am actually getting ready to import the 77 crawl into my Vegas project today and align it to the ships in the 81 crawl from the 35 mm scan and the 85 Widescreen LD (I need a second source because the 35 mm scan has a lot of gateweave issues and a couple of severe jerks. I think I have them accounted for but I like a second source to confirm it). Then I’ll export to png my matte master (that filters out the stars and planet) and a beauty pass of the ships with noise reduction (if I don’t clean it up, when I add the noise and grain later it won’t match the other elements). Then I clean up the matte layer, extracting the frames for the lasers and engine glow before I erase them from the matte. Then I just need to put the layers back together. If my test of adding the engine glow works like I hope, I will use two layers, one for lightness and one for color. The test came out too purple, but I’m not going to try to fix that at this stage.

Author
Time

Sounds complicated, but I think you’ll do a great job of it.

Questions…Will the engine flares of the Star Destroyer, have the same look as seen in Harmy’s DeEd 2.5 - 2.7? Or will it be more, or less? And does the '81 version differ from the '77 version, in the flaring?

Author
Time

The 81 and 77 engine flares are identical (though in pulling them out I could end up with some artifacts that don’t quite line up, but I’m hoping they won’t be noticeable). And no, it will not look like the DeEd. That is more based on the 97 flyover from the BR. It is dressed up to look like the 77 version with the crawl, starfield, and lasers, but the engines and engine glow are form the 97 SE. The 81 and 77 engines and engine glow are distinctly different. For one thing, the engine glow is more blue and cyan on the engines and the flares are completely different. The surrounding engine glow on the stars, planet, and moon is more cobalt and not such a bright blue. I will be lifting the 77 engine glow and using it as intact as I can. I might have to manually do some editing where the moon is because the moon is in different positions in 77 and 81, but my early tests indicate that I might not need to.

I’m having some computer issues so I haven’t been able to access the software to do some batch image conversions. The flyover is 415 frames and a lot of them are going to require manual editing. I already aligned the flyover to the background (I have aligned everything to the 81 35 mm scan which I aligned to the 1985 and 1993 LD versions) so once I have the 3 elements, it should be quick to put it in the project and render it. I still haven’t decided whether I want to use Vegas to cut the matte out for the beauty pass or if I want to do that manually (where I can then clean up the edges as needed) in my photo editing software. excluding the matte, I’m going to use the lasers/engine glow under the beauty pass setting the layer to lighten and then use that same one again using a color option over the beauty pass. I’m actually getting very close to being finished. Just a lot of image manipulation to get the elements isolated.

Author
Time

Looking forward to seeing some samples, IF you decide to post them. Patience is certainly needed here…I can imagine THAT is an understatement!

Author
Time

Yeah. I just determined that I will need to use the mask via the graphics software. That is merging two different images 415 times and then cleaning up the result. Well, 411… the first four frames don’t have any elements to mask. Plus about 120 images (a lot of which have no lasers, just ships) for the lasers and engine glow. I did find an interesting way to isolate just the engine glow and eliminate the star destroyer. But a lot of work ahead. But at least I don’t have to align and resize the frames like I would with the 81 elements. One thing I did discover (not a big thing, but significant) - for the 81 flyover they refilmed the Star Destroyer (or used a different take). It is hard to tell, but the ship isn’t in 100% the same positions. You can tell when the hanger moves across the top of the screen. The edge doesn’t align when you overlay the two different flyovers. They look identical unless you overlay them and catch it. And I thought they just reprocessed it to make a new matte.

Author
Time

yotsuya said:

Yeah. I just determined that I will need to use the mask via the graphics software. That is merging two different images 415 times and then cleaning up the result. Well, 411… the first four frames don’t have any elements to mask. Plus about 120 images (a lot of which have no lasers, just ships) for the lasers and engine glow. I did find an interesting way to isolate just the engine glow and eliminate the star destroyer. But a lot of work ahead. But at least I don’t have to align and resize the frames like I would with the 81 elements. One thing I did discover (not a big thing, but significant) - for the 81 flyover they refilmed the Star Destroyer (or used a different take). It is hard to tell, but the ship isn’t in 100% the same positions. You can tell when the hanger moves across the top of the screen. The edge doesn’t align when you overlay the two different flyovers. They look identical unless you overlay them and catch it. And I thought they just reprocessed it to make a new matte.

That seems pretty important to me. Shouldn’t the reconstruction use the correct take? I know it’s more work, and I’m not trying to step on any toes, but I don’t see the point of a hybrid 77/81 flyover. Maybe I’m just forgetting why the '81 sources can’t be used.

If I had some gum, I’d chew a hole into the sun…

Author
Time
 (Edited)

Well, for the 1981 crawl, we have a 1982 PAL source (widescreen version extracted from horizontally compressed rather than cropped source), the 1985 US Special Widescreen Edition, the Japanese Special Collection, the 1993 Definitive Edition (CAV - later rereleased as the Faces CLV set), and the scanned 35 mm element courtesy of TN1. All of the home video sources are washed out and have heavy smearing (they don’t even come up to the GOUT level of quality). The 35 mm scan has crushed blacks. So bad that there are few visible stars and heavy shadows on the Star Destroyer that shouldn’t be there. That right there would require a better source to repair and the only better sources are the 77 and 97 versions.

Also, as this is a recreation using the best sources possible to get as close to the original as possible, and as the small detail is not one that is readily apparent (I had to overlay the two on top of each other in contrasting colors to even detect it), it will be just one more thing among the long list of points where this will be close but not exact. I’m aiming for as exact as I can make it, but it cannot be perfect. So I have accepted this small little flaw. It is off by all of 3 pixels in the 1920x1080 image. I have stars off by more than that in some frames (I don’t have the skill to match the lens effects to put all the stars in the correct location in every frame). What it comes down to is that the 77 version from 35 mm is closer to the 81 version on 35 mm than any of the LD verions are. The 1985 US SWE is the next best in alignment but it lacks the quality. The other 3 don’t even align properly. And two of the main places that I would need to fix on the 81 35 mm scan are the hanger and the engines. I did not check the 97 (or more properly the DVD/BR version) because the entire engine section and engine glow is wrong and would have to be replaced and I would still have to use the 77 version but have to edit it onto a different version. The 77 version is the closest I can get to the original without sacrificing quality. My goal is to have it look right and feel right and to pass all but the most detailed inspection. I want it to be the best available until a better source is found for the original.

Author
Time

Good luck with it all, and please keep us abreast of your progress!

Author
Time

I was hoping to automate some of the process, but that didn’t work as I’d hoped. Other than that things are going well, just slowly.

Author
Time

yotsuya said:

I was hoping to automate some of the process, but that didn’t work as I’d hoped. Other than that things are going well, just slowly.

Certainly let us know if you find an easy way to separate the elements. My own work recreating this was very exhausting and slow for simply the matter of rotoscoping all the elements.

Author
Time

Darth Lucas said:

yotsuya said:

I was hoping to automate some of the process, but that didn’t work as I’d hoped. Other than that things are going well, just slowly.

Certainly let us know if you find an easy way to separate the elements. My own work recreating this was very exhausting and slow for simply the matter of rotoscoping all the elements.

I did manage to automatically create a matte, but it isn’t very accurate. But I’m going to use a similar process to do it manually. First, I extracted the starfield and planets from some of the early frames of the flyover. I couldn’t use the earliest frames due to some alignment and color issues so the ones I used were about in the middle of the Blockade Runner’s pass and shortly before the star destroyer. I picked 4 frames from two locations and overlayed them to create an average for each location and then cut out the Blockade Runner from one and laid it over the other to fill the hole. Then in Vegas, I used that image to create an overlay layer using the difference setting. That way it would cancel out the stars and planets and leave a very dark background. Then I aligned the entire flyover to that image to maximize how much was cancelled out (I also used a mask to remove the smaller moon from creating an artifact on the star destroyer). Then I exported the resulting sequence to PNG images and ran a salt & pepper filter to remove the last trace of the stars. This leaves a horrible image, but a wonder black area that if selected properly will erase the unwanted area from the original pristine frame. For the later frames where the blue engine glow still remains, I had to isolate the green layer to get a clean enough image to be usable. But it also gave me the solution to isolating the blue engine glow by taking the two images (full color and isolated green) and removing all the lighter areas of the Star Destroyer and leaving only the blue glow. Still, I have to do all the layer operations manually, though I did automate doing the salt and pepper filter and isolating the green (as a b&w image).

Author
Time

Boy…that is really a lot of work. Hope it turns out great, after all that effort!

Author
Time
 (Edited)

I don’t have anything ready, but this is a sample of just how close the 77 and 81 crawls are. I created this to check the laser blasts to be sure that they are identical in both versions. If you look closely you can see that some of the details on the Star Destroyer do not line up, but it is really hard to see much of anything different when you play it at speed. But as you can see, the engine glow lines up nicely. I have the engine glow layer isolated now and am working on the beauty pass layer (I’m creating PNG files with a transparent background for easy composition). Then I need to work on the laser layer (I need to isolate all the laser bolts so I can control the transparency allowing stars to show through like in the original 77 version - then I can use that to fix the blu-ray).

https://vimeo.com/209821415
password: OT

Author
Time

Thanks for the video. I don’t think there’s any difference between the ship elements used. They look nearly identical. They’re aligned a tiny bit differently, maybe, but they look identical. You might be able to match the '77 ships to the '81 by aligning them and warping the image to match the '81. I’m not sure exactly the best way for that for video, but it can be done using the Photoshop alignment tool for individual frames.

If I had some gum, I’d chew a hole into the sun…

Author
Time

That’s why I posted the video. It is really hard to tell. I only noticed when I had it zoomed way in as I was trying to align them. As I said, it is only a few pixels and I’m sure there are other aspects of this project that are going to be off by a few pixels. If I get all the correct elements and composite them right, I’m hoping no will will notice a few tiny things off.

Author
Time

The Blockade Runner engines are giving me headaches. I’m going to have to do layers and replace the blockade runner with a static frame of the engines to get the fade out right Then trim the Star Destroyer so the engines line up right. Also, in examining the engines in such minute detail, I noticed that in the 77 crawl that the elements have gate weave. Williarob’s current 4K77 project has the stars stabilized, but the moon and planet plate as well as the ships move in relation to the stars. I also noticed that in the 77 version during the pan down that the stars and moon/planet elements move the same way they do in the 81 version. It just isn’t as obvious since there aren’t any starts close to them.

Author
Time

I finally have the ships isolated - 415 frames. Still need to tweak the edges a bit, but that layer is largely done. The tweaks only affect where the Star Destroyer crosses each moon and where it moves in front of the Blockade Runner. Now I just have the laser layer to finish. Much easier since most of the frames are blank, but I have to make sure the laser bolts are free of stars and other artifacts and that I don’t miss any. I have the Engine Glow already done. I have a couple of tweaks left to the stars and planets (making sure the edges of the planets are right and the stars are the right intensity) and then I can make a rough cut.

Author
Time

Very excited to see how this comes out. Thanks for the update!

If I had some gum, I’d chew a hole into the sun…

Author
Time

yotsuya said:

I finally have the ships isolated - 415 frames. Still need to tweak the edges a bit, but that layer is largely done. The tweaks only affect where the Star Destroyer crosses each moon and where it moves in front of the Blockade Runner. Now I just have the laser layer to finish. Much easier since most of the frames are blank, but I have to make sure the laser bolts are free of stars and other artifacts and that I don’t miss any. I have the Engine Glow already done. I have a couple of tweaks left to the stars and planets (making sure the edges of the planets are right and the stars are the right intensity) and then I can make a rough cut.

Looking forward to seeing the results of all this hard work!

Author
Time

And the rough assembly of the flyover is done.

https://vimeo.com/214904546
password: OT flyover

I added some grain and gateweave to hide some of the defects and to test how that plugin works. The bad thing about the way I put this together is that the backgorund is absolutely still and the moving elements from the 77 sequence have a bit of gateweave. There are a number of flaws that I have yet to address and a few that I won’t be touching.