I think it’s super cool that the little “shutter” in the new clip caused by removing the beak makes the Sarlaac seem more alive in a strange sort of way.
Yep, practical effects beat CGI every single time.
If you can’t edit the title of this thread, it might be worth starting a new thread (and just let this one roll off the first page). The current title still catches me every time and I already know you’re not Harmy 😉
I’ve replaced the Slovak dubs in my collection with newer dubs, despecialized. The existing Slovak titles-only subtitles in Project Threepio will still work with the new dubs.
I also updated my Ukrainian dub for Star Wars, with better despecialization.
The original Latin American theatrical dub did not voice over the crawl. I’m not sure which crawl they used (Castilian or localized Latino), but at least one of the prints scanned by the 4K77 team was actually a Latino print. I know this only from the audio captured, I don’t think I ever saw the original scanned crawl itself. This may have changed for the later dub made for the Special Editions, which is on Disney+, but I don’t know much about it.
There is no evidence of a Portuguese crawl for the unaltered films. It’s possible one was made for some iteration of the Special Editions. Brazil’s first dub was for TV broadcasts, and I don’t think there was a theatrical run there until the SE. There was never a European Portuguese dub – Portugal is one of the countries where dubbing is very rare except for children’s programming.
I think what you’re describing for #1 and #2 is partway to an SDH subtitle. i.e. cues for things you may not be able to hear clearly. While I’d happily accept SDH subtitles for non-English languages, I’d have to rely entirely on others to do the translation work (the commercial subtitles generally don’t do this). You could use the English SDH subtitles as a template, just finding the SDH cues and translating them/inserting them into the existing subtitles. The character names would have to be localized to the current translation. No universal abbreviation system would work (think about languages that don’t use Latin characters).
There are a couple Uncle Owen exchanges that are just very dense, and rougher on mid-dialogue line breaks than anywhere else in the trilogy. I’m more likely to solve that problem through retiming than anything else, but only if it’s a noticeable improvement and maintains readability. I’ve even dropped one of C-3PO’s lines (“Shutting up, sir”) in that particular exchange in an attempt to give the other lines more space. It’s just… a lot of dialogue in a short period of time. I see what you’re doing, but I’m not sure this is the way to do it, in non-SDH subtitles, at least.
For #3, you’re definitely not the only one who’s interested in this, but this is too much for this project. These subtitles are designed for the English audio only – adding one more supported audio track, with different translations and even different timing (stormtrooper voices don’t always line up at all, because there are no lips to sync), could theoretically double the size of the project, and then double again for another dub. I definitely recognize that if you watch, for example, the Italian dub with Italian subtitles (or the Italian dub with Croatian subtitles, etc), the dialogue won’t match. That’s a line I drew a long time ago for this project, and I’m not willing to step over it. The project goal is to provide the best possible translations of the English dialogue, and that’s all. Dub translations often prioritize things like lip-matching, so it’s possible no “best English translation” subtitle would match in any dub translation. There are language-specific preservations (Krieg der Sterne, Geurre Stellari, etc), and it’d probably make more sense to pair this type of subtitle with those projects. It’s a good idea overall – I know I occasionally watch films both dubbed and subbed in English, and I don’t like when they don’t match, even though I know why.
For #4, I like the thinking outside the box, but no. Too weird 😉
I have new copies of the Indonesian trilogy dubs, which means that the Star Wars dub in no longer incomplete, and I will likely have better-quality audio for the other two films. It’s the same dub, just a better source. The next Project Threepio will include Indonesian titles-only subtitles that go better with these complete dubs.
PM sent.
Welcome back! I think the Despecialized Editions include the DA track for Star Wars (first film only), but I have not heard anything about DA tracks for the other two films, except the Special Editions.
Let me know if you find anything. I still have a dream of recording new descriptive audio narration, which we could then duck into our other excellent-quality audio tracks. Maybe some day…
PM sent.
PM sent.
Sure, where such things are possible, nothing beats a color separation master or a reference print, and that certainly works best. But these things simply aren’t possible with our Star Wars preservations, so I didn’t think that was particularly worth getting into. The best we have is low-fade prints, in the form of low-fade poly, late seventies dye transfer, and in the case of ROTJ, the simply better filmstock of the era (and some lucky showprint finds). Because they’re not perfect color references, they can serve only as starting points, so the question is how do you know where to take the color from there? After all, if you have several prints on the same fadey media all struck from the same master, thirty years later, they will all look different due to differences in storage conditions if nothing else. How do you know how they looked thirty years ago?
DrDre actually has multiple tools at his disposal. One is his famous color matching tool, which as you correctly point out doesn’t have a lot of value here on its own. But another is his color correction tool, which is where the value is (discussion thread about both tools – I see the confusion now, he’s renamed the tools multiple times so it’s hard to keep them straight). Using color theory, the correction tool (or whatever he calls it these days) estimates the original colors of a single faded image (selected for how well it represents color primaries). Then, using the CLUT created from that single frame, you can use the color matching tool to restore the rest of the shot. You then have something very close to where that print actually started before fading.
And yes, it’s not 100% objective science, there is a little art to this as well. Dre does apply small tweaks to his corrections after the fact. Sometimes there might not be a frame within a shot suitable for his color correction tool, so you’d have to estimate. I suspect he also likely adjusts contrast and black levels to match modern audience expectations, for example, or changes things for consistency. But this tool is key to why his colors are generally closer to objectively correct, in my opinion, than most.
New Hindi dubs for ESB and ROTJ are now available in the usual places. Unlike the earlier Hindi dubs, these are complete. Unfortunately, no complete Hindi dub for Star Wars yet.
If anyone has access the the Hotstar streaming service (in India, Indonesia, Malaysia, or Thailand), send me a PM.
It’s okay, I was being diplomatic in my own way, explaining how there can be so many wildly different color corrections for the same film, without outright saying most of them are wrong. Yes, basically there’s DrDre’s color corrections and then there’s subjectively messing around with color, but that’s not really fair to the non-Dre corrections, some of which are pleasant enough to watch in spite of their wrongness.
Well, Tech prints don’t go through as many optical dupe processes, so they’d get less generational “stacked” grain and also less detail loss from duplication. Optical duplication is sort of a weird concept in this digital day and age, but every iteration of the process added grain and reduced fine image detail… and they did it a lot. That’s why commercial releases tend to go back to the original negatives. In the digital world, duplication is a lossless process.
As for the fading, with this sort of film, all colors fade to pink. If you look at a raw scan of a reel from that period, pink is typically the only color left visible. So it’s actually pretty hard to extract multiple colors from an essentially one-color source (there IS other color present, but boy is it not much), and the result can either appear washed out or like those colorized films, depending on whether you don’t go far enough or go too far, or what your expectations are as a viewer.
It’s hard to overstate how much color is lost over a few decades on film from that era. Even Technicolor, which was relatively low-fade, still fades (and was a little screwed up to begin with, as Technicolor was well past its prime at that point). The colors you see in these preservations are reconstructed. Yes, with lots of effort, care, and knowledge about how the films ought to look, but still reconstructed to some degree.
There was color variation on opening day. Technicolor prints (which, as I mentioned earlier were a little screwed up at that time) had a yellowish cast to them that you wouldn’t have seen on an LPP print, and so on.
So basically the color grade of a preservation is in the eyes of the grader. 4K77 has has a lot more people, and a lot more iterations on its color grade, so IMO it looks better. SSE has sort of receded into the background as far as fan attention goes, and its color grade hasn’t been revisited, so IMO it looks worse. Which of them looks more like an opening-day print? Almost certainly one of the more recent 4K77 iterations, but I’d say there’s still quite a lot of room for improvement there, too.
It’s an inexact science, starting out from a point very, very far from where the color really ought to be. It’s a miracle we even get something close.
PM sent.
FWIW, there are two high-profile projects working toward the same goal: OTD83 and Harmy’s Despecialized Edition ROTJ 3.0. It may help people differentiate your project from the others if you give your project an identity that’s clearly different from the rest, and highlight what may be unique about yours. Otherwise I fear your work may be lost in the crowd. Good luck!
Also, FWIW, broadcast masters can have really impressive audio quality, not just compared to their OTA VHS-recorded counterparts, but really objectively good sound. Emphasis on can have good sound though. They have the capacity to sound great, but not all shows really used that capacity.
I always disliked Yub Nub… until I heard what they replaced it with. Now I’m obligated to admit Yub Nub is the best musical finale for this film by a mile.
By the way, to anyone trying to use Project Threepio’s scripts on a non-Windows platform: I’m sorry. My test machine had been a quirky and ancient Linux box, but now that it’s been upgraded, I can see that most of the cross-platform functions had probably been badly broken for some time.
This will definitely be fixed in the next release, as will many minor bugs squashed during the review.
PM sent.
Sorry, I just use the word unicorn to talk about something rare of excellent quality. 4K83’s print was a special print that didn’t go through as many duplication processes as your average projection print. And it was a print with very, very little wear and tear because it hadn’t been shown much. And it hadn’t been abused in any other way – stored in an area that was too damp or warm, and so on. To find a print in this condition over twenty years after all circulating prints were ordered to be destroyed is pretty damn special. It’s like finding a unicorn. It’s a large part of why 4K83 looks so good.
I’m curious, CatBus – how much do you think AI upscaling can improve print scans to get mildly closer to camera negatives quality (in terms of making despecialization more seamless)?
It depends a little on how clever the AI is. The only project I’m really following along these lines is DrDre’s, and it looks like the improvements are modest but noticeable. I’m a fan of AI enhancement, but I’m not sure it could ever really approach camera negatives.
IMO I think we’re stuck with the detail versus authenticity struggle indefinitely. Something pieced together from the best bits of HDTV, BD, and UHD will always outstrip projection print scans on detail, no matter how much we try to enhance them. But that’s the only metric they’re guaranteed to be ahead on. For despecialization, I think we’ll end up removing more detail from the high-detail sources than adding detail to the low-detail sources, to make them blend. And grain matching too, of course.
I’m firmly on the detail side of the struggle. My personal goal is to be able to watch the Star Wars trilogy as if it had had a respectful Blu-ray release (we’re talking 1080p here, I’m not picky), akin to other classic films respectfully restored for the format. Imperfect authenticity isn’t a dealbreaker for me, as long as I can’t detect it.
That said, projection print scans have factors in their favor I can’t ignore. 4K83 looks good. DrDre’s AI enhanced version looks damned good. Does it have all the detail I could ever want? No. But it’s extremely close to good enough to stop caring about it, as far as I’m concerned.
4K77? Meh, I honestly never cared for it. I don’t know much about 4K80, but that ESB print Poita had looked so good it made my teeth hurt. God, I hope that sees the light of day.
I am not familiar with my fair lady’s releases (I have never even seen my fair lady). So I can not understand the comparison, can you please explain for me.
Nothing much to explain – you offered two examples of films that had been butchered or at least abused on UHD. My Fair Lady is just an example of a film that had been treated very respectfully. They still went back to earlier-than-projection-print elements, and so the UHD has more fine detail than the projection prints ever did – and considering it was a 65mm film, that’s really saying something!
I was not saying that all official releases are subpar quality, stuff like the birds and doctor Strangelove or how I learned to stop worrying and love the bomb look beautiful. The camera negatives do show much more detail, just as soon as DNR is in play they show less detail.
Right, but as we see with Despecialized, even if you take a subpar scan of a more detailed source, DNR the crap out of it, and downscale it to 720p, it still can show more image detail than a pristine projection print scanned at 4K. A lot less detail than the negative is often still more detail than a projection print. Yes, for this to be true, the amount of detail lost by generations of optical duplication would have to be staggering – and it is.
Example:
Now don’t get me wrong. The image on the left (Despecialized scaled up to 4K) has all sorts of issues: objectively shitty scan, DNR, edge enhancement, some banding (some of this possibly exaggerated/introduced by the upscale), and of course any fine image detail above 720p is just plain gone. Not perfection in the least, and I could see how a person could reasonably prefer either of the images (because bears). But Despecialized definitely, without any doubt at all, in spite of its many other faults, still has more fine image detail than 4K83 at 4K on the right. And 4K83 is the special unicorn print with much more fine image detail than 4K77.
I am not familiar with my fair lady’s releases (I have never even seen my fair lady). So I can not understand the comparison, can you please explain for me.
Nothing much to explain – you offered two examples of films that had been butchered or at least abused on UHD. My Fair Lady is just an example of a film that had been treated very respectfully. They still went back to earlier-than-projection-print elements, and so the UHD has more fine detail than the projection prints ever did – and considering it was a 65mm film, that’s really saying something!