In my view a restoration in the truest sense of the word would get the highest quality source possible, preferably the original negative or separation masters, repair any damage, remove dirt that was not in the original shot and present the shot as is, warts and all. Which would very likely include a lot of grain, dirt and color noise as Mike points out in his video. However, Mike goes well beyond that using modern digital techniques to get the shot to look better and most notably smoother than it could ever have looked in 1977. Now, how is this different from going back to the original elements and using modern digital techniques, namely digital compositing to get the shot to look better than ever, which aside from the CGI Ronto is exactly what we got with the Special Editions. So, isn’t this shot not just another Special Edition? Additionally, I see a problem in removing grain this way. Namely, the lack of detail is indelibly linked to the grain and overall degradation of the shot. Removing this much grain and noise without significant detail enhancement results in a DNR-ed waxy image, that I’m not sure is a real improvement over the grainy alternative. What do you think?
I should add, that I really like the idea of taking multiple prints and combining them in an effort to get an image much closer to the original negative, but many of the shots Mike discusses go well beyond that.
Verta has an inconsistent approach to Legacy. First, he said at least once that since film negatives degrade even under the best possible storage conditions, it’s not really possible to determine the original state of the negatives at the time they were just put together and the first interpositives could be printed (at best you could do educated guesses, but that’s it), and as such a “purist, perfect restoration” wouldn’t be possible, you will always compromise no matter what. He has also stated that, in his opinion, a true (as best as possible, that’s it) restoration consist of going back to the best available sources and use them to put information back into your restoration. Say, there’s a scratch or tear in this frame, then you use the same frame from another (high quality) source instead of recycling the same area from adjacent frames in the original source.
But then, he goes up and tweaks this scene as you mention, as well as the very rough looking sandcrawler scene (and in the “very same way” as this scene, more on this later). He also remade the opening Star Wars logo and the crawl by combining multiple early frames of each letter/line so that he could create a non-optically and grain limited version of the whole thing. Which would be fine if you don’t care about using “the very original information” as stated earlier so that later in your pipeline you apply some lens distortion and grain-based information loss modeled after the original lenses and film stock used respectively. But here he’s doing the opposite of what he said you ought to do with regards to missing information. And his only justification is that, because he’s using multiple generation sources, he had no choice but to stack early frames together (and then add the original negative grain, and the same optical limitations)… for specific parts of the movie, and not others. And with no hint or clarification as to why only some scenes and not every single shot of the movie that could “benefit” from it. And even he admitted (at least in the case of the crawl) that the intermediate result (the non optically/multi-generation limited crawl) is probably better than how the crawl would have looked when the negatives were still brand new.
In short, he has made a new crawl, that happens to look like the original crawl, from the “same” final frames (albeit taken from lower fidelity copies), but it’s not the original final result, it’s a brand new output, that is claimed to be a “true” restoration.
Also, he has been shown to have lied about the contents of at least his video. The 20th Century Fox logo? He used the logo from one of the Alien blurays. The sandcrawler shot? Apparently he used some good old DNR, because someone tried to use the same method Mike Verta described in the video and couldn’t replicate the results, and because (IIRC, I read this a long time ago) the sources files for the scene that Verta published on his site showed evidence of specific DNR from a certain software having been applied to the entire shot.
I think he also didn’t really state how he dealt with the fact that, when stacking multiple sources together, the lower quality sources would cause the whole thing to lose quality as well. He at least admits in some of his videos that it is a problem, but not how he worked around it.
I also fear, that the focus on maximizing detail and minimizing grain for individual shots will result in a lack of cohesion between shots, because the techniques used may work wonders on one shot, but much less so on another. The result will be a mishmash of shots that are supremely detailed approaching the negative, and others that are only slightly better than the best technicolor prints.
IIRC poita himself said here that he was at Verta’s talk about his restoration in some film preservation conference and the general opinion of everyone there (or at least poita’s opinion) of Legacy was that it was an uneven restoration that faced much, much better on static shots, while the non-static ones still needed a lot of work.
EDIT: just remembered, idk if it’s still true in whatever state Legacy is now, but at some point he said he hated how the Star Wars logo wobbled so much on pre-1979 prints, so he made two versions of it and he was going to use the non-purist one in Legacy while the other would just be for archival purposes. He also made a version of this other scene without the horrible tear: https://vimeo.com/122228097