hairy_hen

Cool.  I didn't expect you'd be doing all of them, but I figured I should point out as much stuff as I could for the sake of being thorough.  Commenting on the colours is of course a dubious prospect without actually having seen the references myself—if what may seem to be errors are in fact part of the original appearance, then that's what it ought to be.

Looks like you're covering the most important stuff, anyway, so I'm looking forward to seeing how those turn out.  I'll make a list for the second half as soon as I can.

People are of course entitled to their own opinions, but that doesn't mean that they're not completely wrong.  ;)

For my part, I really liked the movie.  I've seen it twice now, and look forward to watching it again.  I'm quite a fan of Christopher Nolan's films and his realistic version of Batman, and seeing all three as a triple feature on a 70-foot screen really showed how well they work together as a complete story.  I do admit that the third installment isn't quite as good as the second, which itself isn't quite as good as the first (Batman Begins will always be my favourite), but all of them are extremely enjoyable in their own way.

It seems to me that Selina is the movie's secret weapon.  She's not quite essential to the plot, but her presence makes things a lot more interesting and ties it together in a way that wouldn't have worked as well if she hadn't been included.  The end, in particular, is very compelling from the character development perspective, in no small part due due to this.  Anne Hathaway totally nailed the part, I think.

Harmy, don't go making any final encodes just yet: I'm making a list of remaining errors I've spotted in the workprint.  Though most of them are relatively minor in the scheme of things, fixing them will go a long way towards a seamless presentation of the movie, and would definitely be worthwhile.

Only a very small number of these would go under the category of 'mistakes' in terms of the work on Despecialization.  The majority are problems with the 2004 transfer that have yet to be corrected.  Some of them may be a bit difficult to get right, but I don't think any of them are impossible to manage.  Given all the amazing work that has been done so far, making these last few fixes would take this project about as close to perfection as it could ever get.

So far I've only listed the first half of the movie, but I'll be going through the second half pretty soon.  I've kept comments on the colour timing to a minimum, since for the most part it looks fantastic and I know you're working from the most reliable reference there is, but occasionally issues with the 2004 source creep in and could use some further work.

(Would have posted screenshots, but unfortunately I'm not able to do so, nor to lock in on the exact frame in every instance.  Still, I hope that reference to the onscreen timecode will be helpful.)

00:02:27 — White level too low compared to the two previous shots; inconsistency very noticeable in stars and somewhat in rebel ship.  Brighten image to match, or matte out ship and brighten stars separately if necessary.

00:03:10, 00:03:29 — White level blown out in background for both shots.

00:03:42 — Red, yellow, and blue lights in background extremely oversaturated throughout this shot.  Mike Verta pointed out that they absolutely should not 'pop' out in this way, because it is photographically unrealistic to be so bright through all the obscuring smoke.  However, they must be toned down separately without affecting the laser blasts.

00:03:45 — Red lights in background oversaturated.

00:03:52:05 — Background lights behind stormtroopers again extremely oversaturated and pop unnaturally.  Will again affect laser blasts unless treated separately.

00:03:53:19 — Red lights behind rebels again very oversaturated and photographically unrealistic.  Quite distracting.

00:03:56:11 — Though lasers and flash frames seem appropriately coloured, the explosion on the wall at right of frame still has a very ugly pink appearance.  In the 2004 version it was outright purple, so the removal of magenta throughout most of the movie did not completely succeed in correcting this.

00:04:18:15 — Flash frames throughout this shot are still nasty pink rather than red-orange like they are elsewhere.  (See above comment for explanation.)  Laser blasts also look strange and oversaturated.

00:04:41 — Red reflection on R2 appears strangely 'clipped': flat and without nuance of detail.  Same goes for red light above Leia at 00:04:54.

00:06:47 — Noticeable garbage matte around escape pod.

00:07:07 — Stars disappear around escape pod and are slow to reappear after it has moved on.

00:14:22 — Blue panel on escape pod is missing nearly all of its colour, appearing dull grey instead.

00:24:27 (timing approximate) — Noticeable shift in brightness of the left sun near the beginning of this shot.

This is more of a general comment than specific, but starting around the time Ben finds Luke (around 28 minutes in, and going until the end of the scene), most of the shots still look too desaturated.  I recall the Technicolor image posted by Mike Verta having significantly stronger yellow levels than anything seen here.  Also, Ben's cloak often looks too dark, closer to black than brown.

00:32:02 — I couldn't catch the exact timing, but early in this shot there is a white flash to the right of frame, which looks out of place amid the clean appearance elsewhere.

When Luke ignites the lightsabre, there seems to be a 'halo' around the blade.  Not sure if this is due to compression artefacts or if the GOUT element is interacting strangely with the rest of the image.  I don't recall seeing it before.

00:36:39 — This shot of Vader appears oddly desaturated and dull grey.  In the previous workprint, this shot had a noticeable green cast to it, which is now gone.  Which is correct?

00:39:47 — White levels appear a bit blown out in this shot, particularly in the reflections on the landspeeder.

00:41:51 — There is a lot of frozen grain and/or compression artefacts in the sky in this shot.  The stillness of it is distracting.

00:50:51:01 — Though brief, the pink/purple colouring to this explosion is noticeable.  The rest of the frame isn't so bad, though.  This happens a couple times earlier in the shootout, also, but I couldn't catch the exact timings.

00:54:19 — Death Star's laser blast seems to lean too much towards blue-green in this shot, rather than just being green.  Death Star itself has a slight pink tint, which looks . . . odd.

00:54:25 — Creatures on game board seem oversaturated with pink and purple; inconsistent with other shots.  Would require separate adjustment to avoid affecting lightsabre.

Anything I haven't commented on here, of course, looks absolutely wonderful and impressive.  ;)

I'm very nearly ready to sign off on the new 2-channel version.  Just need to check it a little more to make sure there's nothing I've missed, give it a final listening test, and it should be ready for uploading.  I spent a lot of time recently going over it in fine detail and making little tweaks to fix things as well as I could; I think it's all taken care of, and at any rate whatever remaining issues would almost certainly be beyond my skill to improve any further.  I'll post more in more detail about that later on.

My belated thanks to everyone who PM'd me about the digital '77 stereo track.  I did end up using it for the explosion when the droids are getting in the escape pod, since the '85 mix just sounded too tinny in comparison at that point.  The tradeoff is worthwhile for a more robust sound, seems to me.

I think I'll go ahead and use the LFE from the '97 AC3 laserdisc like before, since the DTS version is not yet available as a source.  But when/if that happens, I'll investigate the possibility of replacing it if it turns out to offer an additional benefit to the sound quality.

I want to have the 5.1 track finished in the next few weeks so that it can be included on the Despecialized Edition 2.0, without causing Harmy any unnecessary delays.  I would be spending today on it, but we're going to see a triple-feature for all three of Christopher Nolan's Batman films at the Ciné Capri, so that's right out.  ;)

Thanks to Mike for weighing in.  I agree with his perspective on things, ultimately.

With the Senator screening showing an example of a Technicolor print that clearly did not have the extreme blue cast in those two shots, and taking into account that no two such prints came out looking completely identical to each other, I think we can safely assume it is an error that was not more widely seen, and not indicative of the way it is supposed to look.  It is an instance in which a lab mistake on a particular print has been immortalized: usually blue is the first thing to go when film fades, but not here, since the dyes do not diminish at all.

I think Harmy's call is the right one: give the shot slightly more of a blue cast as a nod to the way it looks there, but mostly keep it the way he had it before.  The shift was so extreme that the lightsabres were almost completely drowned out, and that cannot possibly have been the intent.

It occurs to me that since the Technicolor prints were made differently than all the rest, there is a possibility that the colour timing is not completely identical.  I recall Mike explaining that rather than going the usual route from an interpositive, they came from a separate master that itself was derived directly from the negative, and since the negative contains no colour timing information, it seems likely that it would have had to be done separately.  In such a case it is probable they used a reliable reference in order to retain the proper look and avoid the creation of something that radically differed from how the movie appeared elsewhere, but some of the observed inconsistency may perhaps be explained by this.  With the saturation levels being bolder than regular Eastman stock, any errors would accordingly be exaggerated and compounded in the process.

Still, potential errors aside it really does look fantastic, and I'm so glad we'll be able to see something that resembles the Technicolor, even if only as an approximation.  In addition to the superior colour process, those prints would have had greater resolution and less dupe grain due to being one generation closer to the negative, so seeing the film projected that way is probably the best it could ever look; but this is a big step towards bringing back the original appearance.  So, Harmy and Mike, big thanks from me for that.

And if this download would ever hurry up, I'd actually be able to check out the latest previews . . .

The top shot looks better, but if the bottom one really represents what it looked like on the Technicolor print, then that's where my vote would have to be, for the sake of accuracy to the original appearance.

It does, however, raise the question of whether the Technicolor version was timed exactly the same way as the regular prints were.  Does Mike Verta have any particular insights about this?

I've been reading the thread about the '77 stereo mix in digital, but I haven't actually heard it (if I am to do so, it would have to be posted somewhere other than newsgroups).  It is definitely a great discovery for the sound quality of that mix, though I don't think it will affect anything I'm doing for the 70mm project.

My main reason for wanting to use the '85 mix in my edits this time instead of the '77 was that its wider stereo image matches that of the '93 mix—in the previous version the reduction in width was (to me, at least) easily detectable at times, and limited my options in choosing where to switch between sources and disguise the transition.  In the new version I've had more freedom in this regard, and often even I can hardly hear the switch once the EQ has been matched.  Some of the most troublesome edits may only be given away by something as subtle as a change in the tape hiss.

Occasionally I have felt that the original stereo mix could be a bit more robust-sounding than the 1985 in brief parts, so I guess I haven't completely ruled out using it a little bit; but if possible it would be better not to, if this discrepancy can be otherwise accounted for.

As a word of warning/disclosure, sharp-eyed viewers may notice that the new 70mm track will briefly drop synch about halfway through the movie (when the TIE fighter sneaks up on the Millennium Falcon at the ruins of Alderaan) before regaining it a few seconds later.  This is because the '85 mix ran longer than the '93 here for some reason, and I couldn't get an exact match without making an audible cut in the middle of dialogue and noticeable sound effects.  I haven't yet tried muxing it to see what it looks like, but since the delay amounts to less than one frame and is of short duration, I don't think it will be much of a problem.  Of course, if I do find a way to fix it without the delay or making a jump-cut, then I will certainly do so.

Earlier I spoke of finding two sound effects in the surround channel which I couldn't positively identify as being 1993 additions or authentic 70mm originals, but just to be safe I deleted them from the track, since the former case seems more likely.  I also found another very subtle effect that was added to the '93 mix and hasn't been pointed out before: a tiny little rumbling sound while the rebels are waiting for the stormtroopers to board the ship, which I think is also in the mono mix but definitely absent from the in-theatre 70mm recording.  It's so quiet I'm not surprised no one ever noticed it before, but for such a tiny addition it sure did take a lot of work to eliminate.  This addition corresponds to where the special edition inserted some very loud effects of the ship being pulled in overhead by the Star Destroyer, and the SE version may actually include this as part of it.

After working on this almost nonstop for the past few days, I'm just about done with the main channel editing.  For the most part it sounds about as good as I could ever hope to make it, and I'm really pleased with how it has turned out.

I was going to upload the stereo track and call it finished, but upon further listening tests I began to detect a few subtle errors that had previously escaped my notice.  So it's not quite ready for posting yet, but it will be once I've got those sorted out.  Unfortunately I'm not entirely sure what the trouble is—I can hear that something is mismatched without quite being able to identify exactly what is causing it, or knowing what to do about it.  Some of the things I tried actually made it worse, so if I can't fix it, I guess I'll have to decide how much aural imperfection I'm willing to tolerate.

These are only small errors I'm talking about, quite possibly things that no one will ever notice if they didn't hear the more audible changes that occasionally appeared in the previous version.  If I'm stuck with them, it's certainly not a big deal; I'd just rather know that they're beyond my ability to deal with before putting it out there.

Recently I've become concerned about the issue of phase shift and how that can negatively affect sound quality.  It's a complex subject that I don't understand very well at all, but I've been trying to learn more in order to be able to avoid or counteract it.  For example, I hadn't realized until now that using a low pass filter causes a phase shift in the extracted low frequencies, which is exactly the kind of thing that could cause unintended cancellation or smearing when combined sounds that are no longer aligned.  See here for more information.

Owing to this, and the previous experience in deriving bass from the 1993 mix as part of the LFE channel, I'm thinking now that I'll probably avoid using the '93 as a source in that way again, and instead use edited versions of the '97 bass in those instances, replicating the earlier effects as closely as possible.  Sometimes the '93 bass really does sound the best, which is why I wanted to use it, but combined with the fact that phasing issues can be equipment-dependent, making it impossible to guarantee that other people will hear the same results that I do, that decision seems like the wisest course.

In other threads I've mentioned that I want to use the LFE channel from the 1997 DTS theatrical audio if possible, to gain a potential improvement in sound quality over the AC3 laserdisc, although if that doesn't happen, it'll still be fine the way it is.  Right now I'm looking into joining a sound design and audio editing program for the fall semester, so I definitely want to have this mix all finished by the time that starts.

Theatrical DTS is a completely different format from home video versions, both in the compression algorithm used and the way the audio data is actually stored.  There is no 'core' track to be found.

There is also not actually a discrete LFE channel in the theatrical format, either; the low frequency information is stored in the surround channels and then low-pass filtered out at 80 hz.  This was due to there being no provision made for channels of less than full frequency range (compared to Dolby Digital which from the start has used a channel of limited range up to 120 hz for the LFE), and insufficient bandwidth to add a sixth full range channel.  The home versions of DTS operate more similarly to Dolby's model in this regard.

Those old SE bootlegs do not contain true DTS tracks, from what I recall hearing.  They are simply the AC3 laserdisc mixes converted to the DTS codec in order to have the name recognition, but there is no actual benefit to them.

I can't say with absolute certainty that's what is really going on, of course; but on my setup, which I'm confident is pretty accurate, it sounds subjectively much too loud in the rear channels, that's for sure.  3 db of gain amounts to about a 40% increase in volume, which is a pretty big difference.  It's not subtle.

The difference in movie theatre calibration arises from the days of mono surround when the entire surround array was calibrated at 85 db to match the front channels, and the introduction of stereo surround meant splitting it in half each at -3 db, to maintain constant acoustic power when the halves were summed together.  This was done to maintain backwards compatibility without having to adjust the levels differently depending on the audio format of an individual film.  Since that calibration standard has remained unchanged (other than for Imax which uses single speakers for each rear channel the way home theatre does), 5.1 mixes are pretty much always going to have their rear channels 3 db higher if they were mixed on a typical recording stage, and thus require the appropriate reduction in level when being transferred to home video.  Though it usually is done properly, mistakes are sometimes made.

Apparently some early laserdiscs with 5.1 sound simply ported over the theatrical mix directly without recalibrating, and would thus have this same kind of problem.  It is also possible that some don't get lowered because someone in charge thinks it actually sounds better that way.  The SE mixes of ESB and RotJ seem very overbearing in the surrounds too, and in that case I suspect the tin ear of a certain Matthew Wood is to blame . . .

Not a chance—at least not as a primary source.  Though the 1997 SE mix of the first movie was mainly derived from an original four-track master, it has so many additions and changes piled on top of it that deleting them all would probably be more trouble than it's worth, and there is no suitable discrete channel source to replace them with.

Worse, a fairly severe level of dynamic range compression was used during the mastering process, supposedly to eliminate 'harshness' in transient peaks, so the broadly powerful sound of the 70mm version has been lost due to this destructive alteration and cannot be recovered.

The SE mixes of the other two movies are completely new and have little resemblance to the way they sounded in 70mm, despite having less changes in content made overall.  The 'aural aesthetic' is not the same.

Emulating the 70mm mixes by using the 1993 versions and editing as needed comes much closer to that original sound, despite having to settle for the upmixing of matrixed stereo tracks in place of true discrete channels.  The SE's make good sources for the bass, but some parts have to be substantially altered in order to sound like they actually belong there, and others deleted altogether.

At any rate, as I've mentioned before I'm definitely interested in using the LFE channel from the DTS track of the first film for the new version of the recreated 70mm mix, as it may provide improved fidelity over the AC3 laserdisc.  As such I eagerly await news of progress.  ;)

Sigh.  Like -1 said, it has nothing to do with censorship.  I'm simply respecting his wishes in not publicly discussing the more open communication his chats are able to have about a subject that, due to the unfortunate reality of things, is in a legally very grey area indeed.

I think some people here have a misplaced sense of entitlement when it comes to these movies, forgetting that for all the unquestionable moral imperative of preservation, no one but LFL actually has any legal right to them whatsoever.  Believe me, I wish that were otherwise as much as anyone, but it isn't.  So when talking about making a film restoration from the kind of materials that Mike has access to, there is no way most of that can be mentioned openly in a public forum.

Of course, those who would rather sit around making cracks and being needlessly spiteful about it all are welcome to it, but don't expect me to sympathize when the target of such 'humour' is doing amazing and necessary work that is probably the best hope this beloved film ever has of being saved from the madness of its owner.

captainsolo said:

Something that's been nagging at me for some time: on LD, which sound track is preferable, Analog or Digital? If there is a Digital track present, is the analog usually sourced from this or is it possible to make an analog transfer of a higher quality bitrate than the encoded 16/44.1 of the Digital track? I've always thought the Digital track was preferable due to being presented discretely and typically cleaner, but after dabbling in LP needledrops and higher resolution audio I wondered about the analog tracks.

I'd love to find the original Raiders LD and make a 24/96 high res track of the original Dolby Stereo.

I'm not an expert on such things, but I recall hearing that the resolution of analogue laserdisc sound is comparable to that of FM radio.  So it's good quality, but definitely not the greatest, with less dynamic range and a lower signal-to-noise ratio than can be found in 16-bit PCM.  Capturing analogue tracks in higher resolution before downsampling gives better results than using CD quality from the start, due to the characteristics of digital audio (by minimizing rounding errors and using dither to mask quantization noise), but don't expect miracles in what you're going to get out of a limited source.

Belbucus' copy of the original stereo mix for Star Wars was done using very high quality analogue equipment, and yet the 1985 version in digital generally sounds better, with less noise and greater fidelity.  True, part of this may be due to whatever remastering was done at the time, but I'm certain the difference in format accounts for part of it.

As always, the quality of the equipment used for capture and playback will have a major role in the end result of what is actually heard.

And according to this thread on TheRaider.net, there is an alternate line when Indy is sliding under the truck to lash his whip.

http://raven.theraider.net/showthread.php?t=21633

EDIT: Here it is. Indeed, Indy says something like "Do they think I'm dumb?" Plus the grunt he makes when flung out is louder, and some of the sound effects are different sounding.http://www.youtube.com/watch?v=kvE5fhsytS8

It looks like the discussion there is about a tv broadcast version, and it's very likely that the sound is a remix and not the original.  I saw an old 35mm print of Raiders a few years ago, and I'm pretty dang certain that line was not present.

As far as the DVD mix is concerned, Neil S. Bulk claims that it was made from the 70mm version, and given how similar most of it is to what I heard on the 35, I'm inclined to believe him.  Whether the 70 ever had split surrounds back then or not I can't say, but since the rear panning is well done and not intrusive I'm really not worried about that.

The 5.1 does have a built-in flaw, however, and it's not one that I've seen pointed out anywhere: namely, that the surround channels are recorded 3 db too loud on the disc.  Do a simple comparison with the alternate language tracks upmixed through Prologic II and the difference becomes instantly apparent; in the stereo tracks the rear portion of the sound is balanced and natural, blending seamlessly with the front, while in the 5.1 it is often overwhelmingly loud and calls far too much attention to itself.  Panning between front and back doesn't work well because the surround portion can actually be distinguished as separate from the rest of the mix, which is very obviously just plain wrong for the sound.

Fortunately, there is an easy solution if you wish to hear Raiders with the proper balance—all you have to do is go into your receiver calibration and change the playback level of the surround speakers, reducing them each by 3 decibels.  Doing this will allow the mix to sound like itself again, and it provides a much more satisfying listening experience for the movie.  Just don't forget to put the levels back to normal when it's over!  ;)

(This assumes, of course, that you have your 5.1 system properly calibrated already, with all channels level-matched to each other at the listening position.  Many people set their surrounds and subwoofer too high, mistakenly believing that they should stand out as much as possible rather than blending into the system as a whole.  Incorrect speaker placement is also a problem—get those surrounds further away and slightly elevated!  How are you being 'surrounded' if the rear speakers are right next to your head?)

As for why the Raiders mix has this problem, I can't say with complete certainty, but I suspect it may be glitch and not a deliberate decision.  Movie theatres are calibrated with the surround channels playing 3 db lower than in home settings, due to using arrays of multiple speakers rather single point-sources, so all surround effects are deliberately recorded 3 db louder as a result of this.  Part of transferring film mixes to disc is supposed to involve lowering the rear channels to compensate, but it's entirely possible that someone forgot to do this and it simply didn't get noticed.  Whether the Bluray will also have this problem, we'll have to see.

I was at Mike's most recent chat on May 25 (my first), and it was definitely worth it.  Without being able to discuss what was said, I'll just mention that I learned a few things that gave me even more enthusiasm and appreciation for what he's doing than I'd already had, and I can say that all aspects of the movie are being preserved and restored in the highest possibly quality.  Mike is doing great work, and for my part I'm definitely looking forward to chatting with him again at some point, as well as hearing more about his progress.  Just knowing the movie is having that level of attention and care paid to it gives me a lot of hope, even with the unfortunate reality of circumstances being what they are.

What most people seem to forget utterly is that Mara Jade actually uses a blue lightsaber.  At the end of The Last Command, Luke gives her his old weapon as a gift, and in Vision of the Future she is stated to still be using the same one.

Since VotF is the grand finale of the Star Wars story (I do not acknowledge the existence of virtually anything that follows), it can be asserted that she still uses the lightsaber that once belonged to Luke's father.

Works that portray her time as the Emperor's Hand do show her using a magenta lightsaber, but following her master's death she did not possess this weapon any longer.  Anything showing her with this blade color that is not of that time period is a continuity error.  It's not altogether surprising that these details are overlooked, since most artwork tends to show her in leather catsuit type outfits, but if you actually bother to read you'll easily discover that this bears no resemblance whatsoever to Zahn's descriptions of her.

Whoa, cinema DTS preservations?  I certainly wasn't expecting that!

I was going to point out the necessity of reducing the surround channels by 3 db and low pass filtering them at 80 hz for the LFE content, but I'm glad to see that has already been taken care of.  I've heard it suggested that the first DTS laserdiscs did not always successfully adapt theatrical mixes to the home environment, because they forget to adjust the surround and LFE levels and simply ported over the versions calibrated for movie theatre acoustics, so it is likely that this will more accurately represent the sound of the original mix than the LD.

Out of curiosity, is the LFE low pass supposed to happen before or after the surround attenuation?  It wouldn't do to have the bass consistently too quiet throughout the film, of course.  ;)

I've heard bad things about the various home video remixes of Jurassic Park, but haven't had any opportunity to study or compare them for myself.  In any case, those who have heard it say the theatrical mix is far superior to any of them, but I'd be very interested in hearing more information about this in specific detail.  I've never understood this incessant need for movies to be remixed for home video aside from minimal recalibration to adapt them to the acoustics of smaller listening environments.  Why shouldn't we be able to hear exactly what the mixers actually heard when they were making the movie?  Makes no sense to me.

About the 1997 Star Wars mix: yes, I would indeed be interested in the DTS to use for my 70mm project.  I've been limited to using the 384 kbps AC3 from laserdisc, so a higher bitrate version would certainly be a good thing, if it can be managed.  I only actually need the LFE and not the main channels, since that's the only part that the SE is actually used for.

The quality of the recording was surprisingly good; better than I had expected.  Unfortunately, it is indeed the 35mm stereo version, so it didn't help me any with my project as I'd hoped.  Clearly the recording was made at some point before the Cine Capri received a 70mm print.  Still, it was very interesting to hear all the audience reactions to the movie.  :)

No can do—if you're referring to the one at 1:28, it's from the '93 at that point, and I'm only adjusting the '85 to match it, not the other way around.  Aside from the alterations, the '93 mix essentially is the 70mm version; the vast majority of it was taken directly from the printmaster and worked on from there.  Even if I could make that sound effect louder, I wouldn't, because the way I see it that's exactly how it would have sounded originally, or as close as I can get given everything I know about it and what sources are available.

The old version used the stereo mix for a longer span, since I wasn't able to edit them as precisely at that point, so it took on its characteristics unavoidably.  The dynamics are flattened both in the sense of peak reduction and of softer elements brought up in relation to the average.  In some ways it might even sound subjectively 'better', but it's less accurate.  Levels of theatre recordings can't entirely be trusted, because they're subject to acoustic and microphone issues as well as tape limitations, so neither the dynamics nor the tonal balance truly reflect what's on the master tape with anywhere near the accuracy that the '93 mix does, and getting as close as possible to the original is the primary goal of this project.

Of course it also has to sound good in its own right, being after all subject to my creative judgement where total accuracy cannot be achieved, and in some ways designed around how I want the movie to sound, but above all it must lean heavily towards what it actually would have sounded like; and if a particular sound effect is reliably determined to be a bit different than might be expected, then that's how it has to be.  ;)

Absolutely.  While the number of main channel edits in the other two movies is far fewer, the benefit of EQ adjustment will surely improve them as well in those parts.  In addition, I'll be able to use the lossless LFE from the Bluray release in all the places where it isn't clipped, gaining a further improvement in fidelity for those with the equipment and ears to notice such a thing.

Neil S. Bulk mentioned there being some kind of phasing issue that popped up briefly in ESB, so I'd like to identify and eliminate this error.  Though I haven't yet heard that, I think I know what would have caused it, so hopefully it will be bypassed altogether.  Since there were not, I believe, any '85 remixes for the later films, narrowed imaging will continue to be unavoidably present in them on occasion, but with the inserts better disguised it shouldn't be too much of a problem.  Nobody but me has ever complained about that, anyway.  ;)

I'd like to make some additional edits to RotJ, because I've noticed that the music is sometimes mixed rather lower in its '93 version than in the original stereo, making it less authentic to the theatrical release.  I already dealt with this during the part where Luke defeats with Vader, but since it also occurs elsewhere I'll probably replace the most problematic instances.  Naturally this would have to be done with as little dynamic sacrifice as possible.

Finally, some news.  ;)

With my S/PDIF connection not working, and having run into some difficulties in achieving everything I'd wanted to do with the new version, I ended up putting it on hold for a while and occupying myself with other things.  But I recently got inspired to work on it again after reading a very interesting article on correcting the perceived sound of headphones in order to achieve an even response: http://www.head-fi.org/t/413900/how-to-equalize-your-headphones-a-tutorial

To do this properly requires the use of a parametric equalizer for precise adjustment of certain frequency ranges, so it was therefore necessary to teach myself how to do this.  Until now I hadn't even realized how much over-emphasis there really was in certain parts of the treble range, and continue to be amazed at the increased midrange clarity I've now gained.  And it's a good thing I made the effort, because it turns out that a parametric equalizer is exactly what I needed to make this 70mm project the best it can be.

The differing EQ of the edited segments was always something that prevented them from integrating as seamlessly as I'd have liked.  Adjusting it would be the obvious answer, but I avoided doing that before since I knew I wasn't truly knowledgeable enough about audio to ensure that I wouldn't make things worse in the process.  I still don't claim to be any kind of expert, but I've at least learned enough to get this thing where I want it.  So I've started going through all the edits again and tweaking the EQ of the '85 mix in order to more closely resemble the '93, and I'm pretty happy with what I've achieved so far.

Getting an exact match between the two isn't possible, or even necessary in some cases, but usually a cut in the vicinity of 2 khz blends them together very well.  Occasionally, more complex adjustment is required, and some of these I've redone more than once to ensure I have the best possible result.  I'm following the principle of looking at EQ through the metaphor of 'sculpting', so that frequencies may only be cut and never boosted (additions can be made only by adjusting the overall gain), in order to avoid adding noise or introducing unnatural resonances.  There seems to be a consensus that digital EQ is inherently inferior to analogue and more likely to create distortion, but since I can't afford a high quality hardware unit I'm doing the best I can with what I have.  So far I haven't heard any flaws in the adjusted sound that weren't already present in the source, so hopefully that won't be an issue.

Using the '85 mix has eliminated the previous problem of narrowed imaging during the replaced segments, and while the peaks are still considerably flattened, I now have greater freedom in choosing edit points, which allows for retention of more of the '93 version, therefore bringing it closer to the 70mm sound.  To illustrate this improvement, I've uploaded a file comparing the new version with the old for the opening shot of the movie: https://rapidshare.com/files/182232409/flyby_comparison.wav

It's not perfect, but it seems to work pretty well.  See if you can tell which is which.  (hint: greater dynamics = good)

I'm about half done with the EQ adjustments, and after that I'll put the finishing touches on the LFE channel.  My S/PDIF issue looks to be solved in the near future, so provided I don't get too busy with unrelated things or lose motivation, I should be closing in on completion fairly soon.

Erikstormtrooper said:

Do you know if Hairy_Hen inserted these 70mm differences in his 70mm re-creation mixes?

I didn't have to insert anything for SW, since my version is derived primarily from the 1993 mix, which itself is mostly a downmix of the 70mm audio to begin with.  Therefore the differences noted for the first film are already present.

However, there are two sound effects which I am at present unable to say for certain whether they were really part of the original or not.  It is likely that they are 1993 additions, but since they appear in the surround channel, and the 70mm recording I have seems to be missing all of its surround effects for some reason, it's impossible to tell.  The first is when R2 gets electrified by the Jawas—a pronounced rolling echo in the rear after the shot is fired.  The other is during the shot of the Millennium Falcon in hyperspace just before they arrive at Alderaan: there is a sort of 'backdraft' sound just before the camera cuts to the inside of the ship.

The echo reappears in the special edition, although mixed slightly differently, but the other is heard only in the 1993 version.  As I said, it's currently impossible to say whether they were added for the laserdisc release, as many other sound effects were, or if they should join the above list of definite changes to the 70mm version from the 35mm.  I'm hopeful that I may be able to clear this up, though.

I suppose the LFE elements of the 70mm mix deserve a mention as well, as they often did not merely amplify the low end of the rest of the track but instead contained new discrete bass content.  Mostly it was used to reinforce explosions and spaceships, but perhaps the most noticeable and unique instance occurs when the ship is caught by the Death Star's tractor beam: there is a pulsing 60 hz tone while they struggle futilely to break away, and interestingly enough this seems to have been among the loudest bass in the entire movie.  (In the SE it is very subdued.)

As for ESB and RotJ, I didn't attempt to duplicate any of the 70mm changes, because I wasn't trying to replicate them exactly; nor did I have adequate sources to do so.  With the second film being edited slightly differently there wasn't any point in that, and for the third I didn't bother because I didn't feel like it, to be honest (nor could I guarantee I'd edit the music the same way).  So those two more properly reflect the 35mm versions in content, but the 70mm in dynamic range.

The other most noticeable difference is that after they get out of the garbage pit and Chewie runs away frightened, the trash monster makes a sort of high-pitched roaring before Han fires a shot back into the door.

This sound is heard in the mono and 70mm versions, but not in the 35mm stereo.  This addition hasn't been pointed out before, to my knowledge.

If this is a good quality recording of the 70mm mix, then I'm extremely interested in hearing it.  I'll be able to tell for sure what it is, I can promise you that.  ;)

A while ago I started to do something like this myself, but didn't finish it for various reasons.

There are many, many differences in the mono mix that have never been pointed out before, and I'm sure there are ones that I haven't noticed, even with all my attention to detail in this kind of thing.

I can tell you right now that putting the mono mix in one channel and a fold-down of the stereo mix in the other won't give you a completely accurate comparison, because all the surround effects in the stereo mix will phase-cancel themselves out, and thus be inaudible.  (And yeah, there are many more surround effects in there than just the noticeable ones, lots of ambience and things of that nature.)  Virtually none of the sound effects that were put in the surround channel were included in the mono mix at all; therefore they'll be altogether absent from both channels of a comparison track and won't register either way.  It just isn't so simple as that, I'm afraid.

The reason for their omission in the mono I'm not sure of, but I would guess that it was probably to reduce noise by having a lesser number of tracks mixed in, since the format was low fidelity and frequency-limited to begin with.

teoalcantara said:

1) Sound seems a little bit distorted, as if volume was a tad too high, but this can be due to mp4 compression or my codecs/player fault (I use MPC-HC);

2) I don't know if it has always been like that (and probably it WAS, due to Special FX technical limitations of the era the film was made), but at 00:56:18:23 and 00:56:58:18 (Luke's saber going off and on) it felt to me like there was a jump cut /missing frame;

The sound is indeed pretty distorted, but what you hear on the workprint is just a temp track.  I've been looking at them with the volume all the way off, anyway.  Be assured that the audio on the final version is of much greater quality than this.  ;)

Those jump cuts in the training scene have always been there.  Even the SE's never did anything about them.  The only place you can see non-jumpy renditions of those shots is in the version Darth Editous put out a few years ago, which used frame morphing to smooth over the transitions with surprisingly good results.