I think this quote needs to be seen again:
"In this case, for all three films, we used interpositive elements that had been made directly from the camera negative. Other film transfers might be done from internegatives made from the interpositive, or from low-contrast prints, but we preferred the IP's for these transfers, because that's the earliest generation usable"
"One small difference from the original films is that in letterbox transfers we prefer to put any subtitles in the black border beneath the actual picture area. Thus we didn't use the same interpositive as the theatrical one, because that one contains subtitling already. In tracking down the elements, we found that the only ones in the vault were ones with subtitles- these clearly weren't the first generation off the camera neg because they had to have the subtitles burned in. So a massive search was undertaken and the first generation IP's were found in a special vault having only opticals in Los Angeles."
"A Mark IIIC with a 4:2:2 digital output [telecine] was used."
"[We used] a noise reduction and dirt concealment device made by Digital Vision, a company in Sweden. Their DVNR-1000 is a very powerful noise reducer for reducing film grain. Especially on the two earlier movies the film grain was very high."
So, let's take this apart based on what is said.
-Previous pre-1993 transfers did not use this "original" IP. They are theorized as using internegatives or print masters, which is probably true. There was also a new IP made for home video in 1985, I believe.
-These are not the theatrical IPs. The theatrical IPs have the English subs burned in.
-All the IPs/prints they could find in the vault had the subs burned in.
-The blank IPs used for the transfer were found in a vault used for storing optical elements in L.A.
So, there are a number of deductions and questions one can extract here. First is this:
-Is this IP actually a different lineage than the "theatrical IP" with subs? If it isn't that means the process of printing Star Wars went like this:
negative>IP (blank)>IN (blank)>IP (subs)>IN (subs)>release print (subs)
That's an awful lot of generational copying! This seems pointless. A more logical way to do it would be this.
negative> IP (blank)
negative> IP (subs)>IN (subs)>release print (subs)
In other words, make two branches. One is intended for theatrical release, with subs. The other is just a straight backup of the negative, blank. You can use it for video, or just as a backup. Maybe this is what foreign prints were made from, using the highly convoluted first process I listed. Maybe this is why foreign prints tend to look a bit rougher, because of the extra two generations? (or is that my imagination/poor storage?)
-It is noted that these IPs were excessively grainy. Is this just a lot of dirt because they were essentially lost in some warehouse in L.A. for a decade and a half?
-If these had been basically lost until they dug them up in 1993--then what the hell were the previous videos being made from? Because they are "blank" as well. One answer may be that they were the INs/print masters derived from this blank IP. All the vault IPs were subbed. But he doesn't say anything about the INs or low-con masters. Maybe they found blank prints that were internegatives or low-contrast print masters, but they were looking for the IP these were made from but couldn't find them, then finally discovered them in a separate warehouse.
-This doesn't answer the question of why they are so grainy. I don't believe it's just dirt, although they are definitely dirty too. Even the guy says grain was a problem, not dirt.
-If the earlier home video transfers used later-generation material derived from this (e.g. the IN and the low-con print derived from that IN)...then why does some of them, notably the 1982 transfer, not have that negative tear (the reason we got on this). If this 1993 transfer is from an original IP made before all the others, why does it have the tear while the 1982 one doesn't, even though that tear was there at some point in 1977? The fact that the 1982 print doesn't have the tear should indicate that that particular print is in fact the earliest one made, as it predates the tear. That print is also blank. So, that would mean that there actually is another blank IP from an earlier time (which also looks less grainy, from the captures).
Based on that, none of this really adds up does it? I do have one theory, which is perhaps unlikely. That theory is that the 1982 blank print is actually the original IP. So where did the blank IP they used come from? No idea. But maybe it was not actually original but was created at a later date from a later IP, which means it actually is the same as a positive release print: original lost IP>IN (used for previous videos)>IP used in 1993.
This would better explain it's grain. But if that's the case...what happened to the original IP? If the only blank material in the vault is INs and print masters, what happened to the IP those were made of? And how convenient that they just happen to find a blank IP hidden at another location (thus giving their story credibility...there was no blank IPs in the vault because it had been stored at the optical warehouse). The only way this would make sense is if that original IP had been junked due to over-use. But if all previous videos were made from the INs etc, that means it was never used. So the only other answer is that it just disappeared. This seems a bit too convenient.
Hell, maybe it's just a case of the films being grainier than anyone realized. This stuff can drive someone mad.
Another worthwhile though is that he says all the "elements" in the vault had subs. By elements does he just mean IPs? If so, why say elements? This seems to imply that every copy of the film in their vault had subs. So again: where did the previous videos get their stuff from? Did these guys just not look in the right place? Wasn't there a record of where the print was for the video release made by the same company only a few years earlier? If they just weren't looking in the right place that would lend credence to the pre-1993-IP=clean-version, 1993-IP=grainy-version theory. But it's hard to believe they didn't have access to the previous film materials. But it kind of seems like he is saying that.